【计算机视觉论文速递】2019-01-01~01-04
- 2019-01-01
- 2019-01-02
- 2019-01-03
- 2019-01-04
本文分享共52篇论文，涉及人脸识别、图像分类、目标检测、语义分割、GAN、姿态估计、风格迁移和文本检测等方向。
[TOC]
《Support Vector Guided Softmax Loss for Face Recognition》
arXiv：https://arxiv.org/abs/1812.11317
Face recognition has witnessed significant progresses due to the advances of deep convolutional neural networks (CNNs), the central challenge of which, is feature discrimination. To address it, one group tries to exploit mining-based strategies (\textit{e.g.}, hard example mining and focal loss) to focus on the informative examples. The other group devotes to designing margin-based loss functions (\textit{e.g.}, angular, additive and additive angular margins) to increase the feature margin from the perspective of ground truth class. Both of them have been well-verified to learn discriminative features. However, they suffer from either the ambiguity of hard examples or the lack of discriminative power of other classes. In this paper, we design a novel loss function, namely support vector guided softmax loss (SV-Softmax), which adaptively emphasizes the mis-classified points (support vectors) to guide the discriminative features learning. So the developed SV-Softmax loss is able to eliminate the ambiguity of hard examples as well as absorb the discriminative power of other classes, and thus results in more discrimiantive features. To the best of our knowledge, this is the first attempt to inherit the advantages of mining-based and margin-based losses into one framework. Experimental results on several benchmarks have demonstrated the effectiveness of our approach over state-of-the-arts.
《Improving Face Anti-Spoofing by 3D Virtual Synthesis》
arXiv：https://arxiv.org/abs/1901.00488
Face anti-spoofing is crucial for the security of face recognition systems. Learning based methods especially deep learning based methods need large-scale training samples to reduce overfitting. However, acquiring spoof data is very expensive since the live faces should be re-printed and re-captured in many views. In this paper, we present a method to synthesize virtual spoof data in 3D space to alleviate this problem. Specifically, we consider a printed photo as a flat surface and mesh it into a 3D object, which is then randomly bent and rotated in 3D space. Afterward, the transformed 3D photo is rendered through perspective projection as a virtual sample. The synthetic virtual samples can significantly boost the anti-spoofing performance when combined with a proposed data balancing strategy. Our promising results open up new possibilities for advancing face anti-spoofing using cheap and large-scale synthetic data.
《Face Recognition: A Novel Multi-Level Taxonomy based Survey》
arXiv：https://arxiv.org/abs/1901.00713
In a world where security issues have been gaining growing importance, face recognition systems have attracted increasing attention in multiple application areas, ranging from forensics and surveillance to commerce and entertainment. To help understanding the landscape and abstraction levels relevant for face recognition systems, face recognition taxonomies allow a deeper dissection and comparison of the existing solutions. This paper proposes a new, more encompassing and richer multi-level face recognition taxonomy, facilitating the organization and categorization of available and emerging face recognition solutions; this taxonomy may also guide researchers in the development of more efficient face recognition solutions. The proposed multi-level taxonomy considers levels related to the face structure, feature support and feature extraction approach. Following the proposed taxonomy, a comprehensive survey of representative face recognition solutions is presented. The paper concludes with a discussion on current algorithmic and application related challenges which may define future research directions for face recognition.
《Multi-class Classification without Multi-class Labels》
ICLR 2019
arXiv：https://arxiv.org/abs/1901.00544
This work presents a new strategy for multi-class classification that requires no class-specific labels, but instead leverages pairwise similarity between examples, which is a weaker form of annotation. The proposed method, meta classification learning, optimizes a binary classifier for pairwise similarity prediction and through this process learns a multi-class classifier as a submodule. We formulate this approach, present a probabilistic graphical model for it, and derive a surprisingly simple loss function that can be used to learn neural network-based models. We then demonstrate that this same framework generalizes to the supervised, unsupervised cross-task, and semi-supervised settings. Our method is evaluated against state of the art in all three learning paradigms and shows a superior or comparable accuracy, providing evidence that learning multi-class classification without multi-class labels is a viable learning option.
《Large-Scale Object Detection of Images from Network Cameras in Variable Ambient Lighting Conditions》
MIPR 2019
arXiv：https://arxiv.org/abs/1812.11901
Computer vision relies on labeled datasets for training and evaluation in detecting and recognizing objects. The popular computer vision program, YOLO ("You Only Look Once"), has been shown to accurately detect objects in many major image datasets. However, the images found in those datasets, are independent of one another and cannot be used to test YOLO's consistency at detecting the same object as its environment (e.g. ambient lighting) changes. This paper describes a novel effort to evaluate YOLO's consistency for large-scale applications. It does so by working (a) at large scale and (b) by using consecutive images from a curated network of public video cameras deployed in a variety of real-world situations, including traffic intersections, national parks, shopping malls, university campuses, etc. We specifically examine YOLO's ability to detect objects in different scenarios (e.g., daytime vs. night), leveraging the cameras' ability to rapidly retrieve many successive images for evaluating detection consistency. Using our camera network and advanced computing resources (supercomputers), we analyzed more than 5 million images captured by 140 network cameras in 24 hours. Compared with labels marked by humans (considered as "ground truth"), YOLO struggles to consistently detect the same humans and cars as their positions change from one frame to the next; it also struggles to detect objects at night time. Our findings suggest that state-of-the art vision solutions should be trained by data from network camera with contextual information before they can be deployed in applications that demand high consistency on object detection.
《Impact of Ground Truth Annotation Quality on Performance of Semantic Image Segmentation of Traffic Conditions》
arXiv：https://arxiv.org/abs/1901.00001
Preparation of high-quality datasets for the urban scene understanding is a labor-intensive task, especially, for datasets designed for the autonomous driving applications. The application of the coarse ground truth (GT) annotations of these datasets without detriment to the accuracy of semantic image segmentation (by the mean intersection over union - mIoU) could simplify and speedup the dataset preparation and model fine tuning before its practical application. Here the results of the comparative analysis for semantic segmentation accuracy obtained by PSPNet deep learning architecture are presented for fine and coarse annotated images from Cityscapes dataset. Two scenarios were investigated: scenario 1 - the fine GT images for training and prediction, and scenario 2 - the fine GT images for training and the coarse GT images for prediction. The obtained results demonstrated that for the most important classes the mean accuracy values of semantic image segmentation for coarse GT annotations are higher than for the fine GT ones, and the standard deviation values are vice versa. It means that for some applications some unimportant classes can be excluded and the model can be tuned further for some classes and specific regions on the coarse GT dataset without loss of the accuracy even. Moreover, this opens the perspectives to use deep neural networks for the preparation of such coarse GT datasets.
《Flow Based Self-supervised Pixel Embedding for Image Segmentation》
arXiv：https://arxiv.org/abs/1901.00520
We propose a new self-supervised approach to image feature learning from motion cue. This new approach leverages recent advances in deep learning in two directions: 1) the success of training deep neural network in estimating optical flow in real data using synthetic flow data; and 2) emerging work in learning image features from motion cues, such as optical flow. Building on these, we demonstrate that image features can be learned in self-supervision by first training an optical flow estimator with synthetic flow data, and then learning image features from the estimated flows in real motion data. We demonstrate and evaluate this approach on an image segmentation task. Using the learned image feature representation, the network performs significantly better than the ones trained from scratch in few-shot segmentation tasks.
《SiamRPN++: Evolution of Siamese Visual Tracking with Very Deep Networks》
arXiv：https://arxiv.org/abs/1812.11703
Siamese network based trackers formulate tracking as convolutional feature cross-correlation between target template and searching region. However, Siamese trackers still have accuracy gap compared with state-of-the-art algorithms and they cannot take advantage of feature from deep networks, such as ResNet-50 or deeper. In this work we prove the core reason comes from the lack of strict translation invariance. By comprehensive theoretical analysis and experimental validations, we break this restriction through a simple yet effective spatial aware sampling strategy and successfully train a ResNet-driven Siamese tracker with significant performance gain. Moreover, we propose a new model architecture to perform depth-wise and layer-wise aggregations, which not only further improves the accuracy but also reduces the model size. We conduct extensive ablation studies to demonstrate the effectiveness of the proposed tracker, which obtains currently the best results on four large tracking benchmarks, including OTB2015, VOT2018, UAV123, and LaSOT. Our model will be released to facilitate further studies based on this problem.
《Generating Multiple Objects at Spatially Distinct Locations》
ICLR 2019
arXiv：https://arxiv.org/abs/1901.00686
Recent improvements to Generative Adversarial Networks (GANs) have made it possible to generate realistic images in high resolution based on natural language descriptions such as image captions. Furthermore, conditional GANs allow us to control the image generation process through labels or even natural language descriptions. However, fine-grained control of the image layout, i.e. where in the image specific objects should be located, is still difficult to achieve. This is especially true for images that should contain multiple distinct objects at different spatial locations. We introduce a new approach which allows us to control the location of arbitrarily many objects within an image by adding an object pathway to both the generator and the discriminator. Our approach does not need a detailed semantic layout but only bounding boxes and the respective labels of the desired objects are needed. The object pathway focuses solely on the individual objects and is iteratively applied at the locations specified by the bounding boxes. The global pathway focuses on the image background and the general image layout. We perform experiments on the Multi-MNIST, CLEVR, and the more complex MS-COCO data set. Our experiments show that through the use of the object pathway we can control object locations within images and can model complex scenes with multiple objects at various locations. We further show that the object pathway focuses on the individual objects and learns features relevant for these, while the global pathway focuses on global image characteristics and the image background.
《Fast and Globally Optimal Rigid Registration of 3D Point Sets by Transformation Decomposition》
arXiv：https://arxiv.org/abs/1812.11307
The rigid registration of two 3D point sets is a fundamental problem in computer vision. The current trend is to solve this problem globally using the BnB optimization framework. However, the existing global methods are slow for two main reasons: the computational complexity of BnB is exponential to the problem dimensionality (which is six for 3D rigid registration), and the bound evaluation used in BnB is inefficient. In this paper, we propose two techniques to address these problems. First, we introduce the idea of translation invariant vectors, which allows us to decompose the search of a 6D rigid transformation into a search of 3D rotation followed by a search of 3D translation, each of which is solved by a separate BnB algorithm. This transformation decomposition reduces the problem dimensionality of BnB algorithms and substantially improves its efficiency. Then, we propose a new data structure, named 3D Integral Volume, to accelerate the bound evaluation in both BnB algorithms. By combining these two techniques, we implement an efficient algorithm for rigid registration of 3D point sets. Extensive experiments on both synthetic and real data show that the proposed algorithm is three orders of magnitude faster than the existing state-of-the-art global methods.
《Skeleton Transformer Networks: 3D Human Pose and Skinned Mesh from Single RGB Image》
ACCV
arXiv：https://arxiv.org/abs/1812.11328
In this paper, we present Skeleton Transformer Networks (SkeletonNet), an end-to-end framework that can predict not only 3D joint positions but also 3D angular pose (bone rotations) of a human skeleton from a single color image. This in turn allows us to generate skinned mesh animations. Here, we propose a two-step regression approach. The first step regresses bone rotations in order to obtain an initial solution by considering skeleton structure. The second step performs refinement based on heatmap regressor using a 3D pose representation called cross heatmap which stacks heatmaps of xy and zy coordinates. By training the network using the proposed 3D human pose dataset that is comprised of images annotated with 3D skeletal angular poses, we showed that SkeletonNet can predict a full 3D human pose (joint positions and bone rotations) from a single image in-the-wild.
《Learning Generalizable Physical Dynamics of 3D Rigid Objects》
arXiv：https://arxiv.org/abs/1901.00466
Humans have a remarkable ability to predict the effect of physical interactions on the dynamics of objects. Endowing machines with this ability would allow important applications in areas like robotics and autonomous vehicles. In this work, we focus on predicting the dynamics of 3D rigid objects, in particular an object's final resting position and total rotation when subjected to an impulsive force. Different from previous work, our approach is capable of generalizing to unseen object shapes - an important requirement for real-world applications. To achieve this, we represent object shape as a 3D point cloud that is used as input to a neural network, making our approach agnostic to appearance variation. The design of our network is informed by an understanding of physical laws. We train our model with data from a physics engine that simulates the dynamics of a large number of shapes. Experiments show that we can accurately predict the resting position and total rotation for unseen object geometries.
《GeoNet: Deep Geodesic Networks for Point Cloud Analysis》
arXiv：https://arxiv.org/abs/1901.00680
Surface-based geodesic topology provides strong cues for object semantic analysis and geometric modeling. However, such connectivity information is lost in point clouds. Thus we introduce GeoNet, the first deep learning architecture trained to model the intrinsic structure of surfaces represented as point clouds. To demonstrate the applicability of learned geodesic-aware representations, we propose fusion schemes which use GeoNet in conjunction with other baseline or backbone networks, such as PU-Net and PointNet++, for down-stream point cloud analysis. Our method improves the state-of-the-art on multiple representative tasks that can benefit from understandings of the underlying surface topology, including point upsampling, normal estimation, mesh reconstruction and non-rigid shape classification.
《Detecting Text in the Wild with Deep Character Embedding Network》
ACCV 2018
arXiv：https://arxiv.org/abs/1901.00363
Most text detection methods hypothesize texts are horizontal or multi-oriented and thus define quadrangles as the basic detection unit. However, text in the wild is usually perspectively distorted or curved, which can not be easily tackled by existing approaches. In this paper, we propose a deep character embedding network (CENet) which simultaneously predicts the bounding boxes of characters and their embedding vectors, thus making text detection a simple clustering task in the character embedding space. The proposed method does not require strong assumptions of forming a straight line on general text detection, which provides flexibility on arbitrarily curved or perspectively distorted text. For character detection task, a dense prediction subnetwork is designed to obtain the confidence score and bounding boxes of characters. For character embedding task, a subnet is trained with contrastive loss to project detected characters into embedding space. The two tasks share a backbone CNN from which the multi-scale feature maps are extracted. The final text regions can be easily achieved by a thresholding process on character confidence and embedding distance of character pairs. We evaluated our method on ICDAR13, ICDAR15, MSRA-TD500, and Total-Text. The proposed method achieves state-of-the-art or comparable performance on all these datasets, and shows substantial improvement in the irregular-text datasets, i.e. Total-Text.
《Accurate, Data-Efficient, Unconstrained Text Recognition with Convolutional Neural Networks》
arXiv：https://arxiv.org/abs/1812.11894
Unconstrained text recognition is an important computer vision task, featuring a wide variety of different sub-tasks, each with its own set of challenges. One of the biggest promises of deep neural networks has been the convergence and automation of feature extractors from input raw signals, allowing for the highest possible performance with minimum required domain knowledge. To this end, we propose a data-efficient, end-to-end neural network model for generic, unconstrained text recognition. In our proposed architecture we strive for simplicity and efficiency without sacrificing recognition accuracy. Our proposed architecture is a fully convolutional network without any recurrent connections trained with the CTC loss function. Thus it operates on arbitrary input sizes and produces strings of arbitrary length in a very efficient and parallelizable manner. We show the generality and superiority of our proposed text recognition architecture by achieving state of the art results on seven public benchmark datasets, covering a wide spectrum of text recognition tasks, namely: Handwriting Recognition, CAPTCHA recognition, OCR, License Plate Recognition, and Scene Text Recognition. Our proposed architecture has won the ICFHR2018 Competition on Automated Text Recognition on a READ Dataset.
《A High-Performance CNN Method for Offline Handwritten Chinese Character Recognition and Visualization》
arXiv：https://arxiv.org/abs/1812.11489
Recent researches introduced fast, compact and efficient convolutional neural networks (CNNs) for offline handwritten Chinese character recognition (HCCR). However, many of them did not address the problem of the network interpretability. We propose a new architecture of a deep CNN with a high recognition performance which is capable of learning deep features for visualization. A special characteristic of our model is the bottleneck layers which enable us to retain its expressiveness while reducing the number of multiply-accumulate operations and the required storage. We introduce a modification of global weighted average pooling (GWAP) - global weighted output average pooling (GWOAP). This paper demonstrates how they allow us to calculate class activation maps (CAMs) in order to indicate the most relevant input character image regions used by our CNN to identify a certain class. Evaluating on the ICDAR-2013 offline HCCR competition dataset, we show that our model enables a relative 0.83% error reduction having 49% fewer parameters and the same computational cost compared to the current state-of-the-art single-network method trained only on handwritten data. Our solution outperforms even recent residual learning approaches.
《Lipi Gnani - A Versatile OCR for Documents in any Language Printed in Kannada Script》
submitted to ACM Transactions
arXiv：https://arxiv.org/abs/1901.00413
A Kannada OCR, named Lipi Gnani, has been designed and developed from scratch, with the motivation of it being able to convert printed text or poetry in Kannada script, without any restriction on vocabulary. The training and test sets have been collected from over 35 books published between the period 1970 to 2002, and this includes books written in Halegannada and pages containing Sanskrit slokas written in Kannada script. The coverage of the OCR is nearly complete in the sense that it recognizes all the punctuation marks, special symbols, Indo-Arabic and Kannada numerals and also the interspersed English words. Several minor and major original contributions have been done in developing this OCR at the different processing stages such as binarization, line and character segmentation, recognition and Unicode mapping. This has created a Kannada OCR that performs as good as, and in some cases, better than the Google's Tesseract OCR, as shown by the results. To the knowledge of the authors, this is the maiden report of a complete Kannada OCR, handling all the issues involved. Currently, there is no dictionary based postprocessing, and the obtained results are due solely to the recognition process. Four benchmark test databases containing scanned pages from books in Kannada, Sanskrit, Konkani and Tulu languages, but all of them printed in Kannada script, have been created. The word level recognition accuracy of Lipi Gnani is 4% higher on the Kannada dataset than that of Google's Tesseract OCR, 8% higher on the datasets of Tulu and Sanskrit, and 25% higher on the Konkani dataset.
《Handwritten Indic Character Recognition using Capsule Networks》
ASPCON 2018
arXiv：https://arxiv.org/abs/1901.00166
Convolutional neural networks(CNNs) has become one of the primary algorithms for various computer vision tasks. Handwritten character recognition is a typical example of such task that has also attracted attention. CNN architectures such as LeNet and AlexNet have become very prominent over the last two decades however the spatial invariance of the different kernels has been a prominent issue till now. With the introduction of capsule networks, kernels can work together in consensus with one another with the help of dynamic routing, that combines individual opinions of multiple groups of kernels called capsules to employ equivariance among kernels. In the current work, we have implemented capsule network on handwritten Indic digits and character datasets to show its superiority over networks like LeNet. Furthermore, it has also been shown that they can boost the performance of other networks like LeNet and AlexNet.
《Image Super-Resolution via RL-CSC: When Residual Learning Meets Convolutional Sparse Coding》
arXiv：https://arxiv.org/abs/1812.11950
github：https://github.com/axzml/RL-CSC
We propose a simple yet effective model for Single Image Super-Resolution (SISR), by combining the merits of Residual Learning and Convolutional Sparse Coding (RL-CSC). Our model is inspired by the Learned Iterative Shrinkage-Threshold Algorithm (LISTA). We extend LISTA to its convolutional version and build the main part of our model by strictly following the convolutional form, which improves the network's interpretability. Specifically, the convolutional sparse codings of input feature maps are learned in a recursive manner, and high-frequency information can be recovered from these CSCs. More importantly, residual learning is applied to alleviate the training difficulty when the network goes deeper. Extensive experiments on benchmark datasets demonstrate the effectiveness of our method. RL-CSC (30 layers) outperforms several recent state-of-the-arts, e.g., DRRN (52 layers) and MemNet (80 layers) in both accuracy and visual qualities. Codes and more results are available at https://github.com/axzml/RL-CSC.
《High Quality Monocular Depth Estimation via Transfer Learning》
arXiv：https://arxiv.org/abs/1812.11941
Accurate depth estimation from images is a fundamental task in many applications including scene understanding and reconstruction. Existing solutions for depth estimation often produce blurry approximations of low resolution. This paper presents a convolutional neural network for computing a high-resolution depth map given a single RGB image with the help of transfer learning. Following a standard encoder-decoder architecture, we leverage features extracted using high performing pre-trained networks when initializing our encoder along with augmentation and training strategies that lead to more accurate results. We show how, even for a very simple decoder, our method is able to achieve detailed high-resolution depth maps. Our network, with fewer parameters and training iterations, outperforms state-of-the-art on two datasets and also produces qualitatively better results that capture object boundaries more faithfully. Code and corresponding pre-trained weights are made publicly available.
《Unsupervised monocular stereo matching》
arXiv：https://arxiv.org/abs/1812.11671
At present, deep learning has been applied more and more in monocular image depth estimation and has shown promising results. The current more ideal method for monocular depth estimation is the supervised learning based on ground truth depth, but this method requires an abundance of expensive ground truth depth as the supervised labels. Therefore, researchers began to work on unsupervised depth estimation methods. Although the accuracy of unsupervised depth estimation method is still lower than that of supervised method, it is a promising research direction.
In this paper, Based on the experimental results that the stereo matching models outperforms monocular depth estimation models under the same unsupervised depth estimation model, we proposed an unsupervised monocular vision stereo matching method. In order to achieve the monocular stereo matching, we constructed two unsupervised deep convolution network models, one was to reconstruct the right view from the left view, and the other was to estimate the depth map using the reconstructed right view and the original left view. The two network models are piped together during the test phase. The output results of this method outperforms the current mainstream unsupervised depth estimation method in the challenging KITTI dataset.
《Epipolar Geometry based Learning of Multi-view Depth and Ego-Motion from Monocular Sequences》
ICVGIP 2018 Best Pape
arXiv：https://arxiv.org/abs/1812.11922
Deep approaches to predict monocular depth and ego-motion have grown in recent years due to their ability to produce dense depth from monocular images. The main idea behind them is to optimize the photometric consistency over image sequences by warping one view into another, similar to direct visual odometry methods. One major drawback is that these methods infer depth from a single view, which might not effectively capture the relation between pixels. Moreover, simply minimizing the photometric loss does not ensure proper pixel correspondences, which is a key factor for accurate depth and pose estimations.
In contrast, we propose a 2-view depth network to infer the scene depth from consecutive frames, thereby learning inter-pixel relationships. To ensure better correspondences, thereby better geometric understanding, we propose incorporating epipolar constraints to make the learning more geometrically sound. We use the Essential matrix obtained using Nist'er's Five Point Algorithm, to enforce meaningful geometric constraints, rather than using it as training labels. This allows us to use lesser no. of trainable parameters compared to state-of-the-art methods. The proposed method results in better depth images and pose estimates, which capture the scene structure and motion in a better way. Such a geometrically constrained learning performs successfully even in cases where simply minimizing the photometric error would fail.
《EANet: Enhancing Alignment for Cross-Domain Person Re-identification》
arXiv：https://arxiv.org/abs/1812.11369
github：https://github.com/huanghoujing/EANet
Person re-identification (ReID) has achieved significant improvement under the single-domain setting. However, directly exploiting a model to new domains is always faced with huge performance drop, and adapting the model to new domains without target-domain identity labels is still challenging. In this paper, we address cross-domain ReID and make contributions for both model generalization and adaptation. First, we propose Part Aligned Pooling (PAP) that brings significant improvement for cross-domain testing. Second, we design a Part Segmentation (PS) constraint over ReID feature to enhance alignment and improve model generalization. Finally, we show that applying our PS constraint to unlabeled target domain images serves as effective domain adaptation. We conduct extensive experiments between three large datasets, Market1501, CUHK03 and DukeMTMC-reID. Our model achieves state-of-the-art performance under both source-domain and cross-domain settings. For completeness, we also demonstrate the complementarity of our model to existing domain adaptation methods. The code is available at https://github.com/huanghoujing/EANet.
《CamLoc: Pedestrian Location Detection from Pose Estimation on Resource-constrained Smart-cameras》
arXiv：https://arxiv.org/abs/1812.11209
Recent advancements in energy-efficient hardware technology is driving the exponential growth we are experiencing in the Internet of Things (IoT) space, with more pervasive computations being performed near to data generation sources. A range of intelligent devices and applications performing local detection is emerging (activity recognition, fitness monitoring, etc.) bringing with them obvious advantages such as reducing detection latency for improved interaction with devices and safeguarding user data by not leaving the device. Video processing holds utility for many emerging applications and data labelling in the IoT space. However, performing this video processing with deep neural networks at the edge of the Internet is not trivial. In this paper we show that pedestrian location estimation using deep neural networks is achievable on fixed cameras with limited compute resources. Our approach uses pose estimation from key body points detection to extend pedestrian skeleton when whole body not in image (occluded by obstacles or partially outside of frame), which achieves better location estimation performance (infrence time and memory footprint) compared to fitting a bounding box over pedestrian and scaling. We collect a sizable dataset comprising of over 2100 frames in videos from one and two surveillance cameras pointing from different angles at the scene, and annotate each frame with the exact position of person in image, in 42 different scenarios of activity and occlusion. We compare our pose estimation based location detection with a popular detection algorithm, YOLOv2, for overlapping bounding-box generation, our solution achieving faster inference time (15x speedup) at half the memory footprint, within resource capabilities on embedded devices, which demonstrate that CamLoc is an efficient solution for location estimation in videos on smart-cameras.
《Rethinking on Multi-Stage Networks for Human Pose Estimation》
arXiv：https://arxiv.org/abs/1901.00148
注：Face++出品
Existing pose estimation approaches can be categorized into single-stage and multi-stage methods. While a multi-stage architecture is seemingly more suitable for the task, the performance of current multi-stage methods is not as competitive as single-stage ones. This work studies this issue. We argue that the current unsatisfactory performance comes from various insufficient design in current methods. We propose several improvements on the architecture design, feature flow, and loss function. The resulting multi-stage network outperforms all previous works and obtains the best performance on COCO keypoint challenge 2018. The source code will be released.
《PVNet: Pixel-wise Voting Network for 6DoF Pose Estimation》
arXiv：https://arxiv.org/abs/1812.11788
homepage：https://zju-3dv.github.io/pvnet/
This paper addresses the challenge of 6DoF pose estimation from a single RGB image under severe occlusion or truncation. Many recent works have shown that a two-stage approach, which first detects keypoints and then solves a Perspective-n-Point (PnP) problem for pose estimation, achieves remarkable performance. However, most of these methods only localize a set of sparse keypoints by regressing their image coordinates or heatmaps, which are sensitive to occlusion and truncation. Instead, we introduce a Pixel-wise Voting Network (PVNet) to regress pixel-wise unit vectors pointing to the keypoints and use these vectors to vote for keypoint locations using RANSAC. This creates a flexible representation for localizing occluded or truncated keypoints. Another important feature of this representation is that it provides uncertainties of keypoint locations that can be further leveraged by the PnP solver. Experiments show that the proposed approach outperforms the state of the art on the LINEMOD, Occlusion LINEMOD and YCB-Video datasets by a large margin, while being efficient for real-time pose estimation. We further create a Truncation LINEMOD dataset to validate the robustness of our approach against truncation. The code will be avaliable at https://zju-3dv.github.io/pvnet/.
《Ancient Painting to Natural Image: A New Solution for Painting Processing》
WACV 2019
arXiv：https://arxiv.org/abs/1901.00224
Collecting a large-scale and well-annotated dataset for image processing has become a common practice in computer vision. However, in the ancient painting area, this task is not practical as the number of paintings is limited and their style is greatly diverse. We, therefore, propose a novel solution for the problems that come with ancient painting processing. This is to use domain transfer to convert ancient paintings to photo-realistic natural images. By doing so, the ancient painting processing problems become natural image processing problems and models trained on natural images can be directly applied to the transferred paintings. Specifically, we focus on Chinese ancient flower, bird and landscape paintings in this work. A novel Domain Style Transfer Network (DSTN) is proposed to transfer ancient paintings to natural images which employ a compound loss to ensure that the transferred paintings still maintain the color composition and content of the input paintings. The experiment results show that the transferred paintings generated by the DSTN have a better performance in both the human perceptual test and other image processing tasks than other state-of-art methods, indicating the authenticity of the transferred paintings and the superiority of the proposed method.
《Interest Point Detection based on Adaptive Ternary Coding》
arXiv：https://arxiv.org/abs/1901.00031
In this paper, an adaptive pixel ternary coding mechanism is proposed and a contrast invariant and noise resistant interest point detector is developed on the basis of this mechanism. Every pixel in a local region is adaptively encoded into one of the three statuses: bright, uncertain and dark. The blob significance of the local region is measured by the spatial distribution of the bright and dark pixels. Interest points are extracted from this blob significance measurement. By labeling the statuses of ternary bright, uncertain, and dark, the proposed detector shows more robustness to image noise and quantization errors. Moreover, the adaptive strategy for the ternary cording, which relies on two thresholds that automatically converge to the median of the local region in measurement, enables this coding to be insensitive to the image local contrast. As a result, the proposed detector is invariant to illumination changes. The state-of-the-art results are achieved on the standard datasets, and also in the face recognition application.
《DCI: Discriminative and Contrast Invertible Descriptor》
arXiv：https://arxiv.org/abs/1901.00027
Local feature descriptors have been widely used in fine-grained visual object search thanks to their robustness in scale and rotation variation and cluttered background. However, the performance of such descriptors drops under severe illumination changes. In this paper, we proposed a Discriminative and Contrast Invertible (DCI) local feature descriptor. In order to increase the discriminative ability of the descriptor under illumination changes, a Laplace gradient based histogram is proposed. A robust contrast flipping estimate is proposed based on the divergence of a local region. Experiments on fine-grained object recognition and retrieval applications demonstrate the superior performance of DCI descriptor to others.
《Mask-aware networks for crowd counting》
arXiv：https://arxiv.org/abs/1901.00039
Crowd counting problem aims to count the number of objects within an image or a frame in the videos and is usually solved by estimating the density map generated from the object location annotations. The values in the density map, by nature, take two possible states: zero indicating no object around, a non-zero value indicating the existence of objects and the value denoting the local object density. In contrast to traditional methods which do not differentiate the density prediction of these two states, we propose to use a dedicated network branch to predict the object/non-object mask and then combine its prediction with the input image to produce the density map. Our rationale is that the mask prediction could be better modeled as a binary segmentation problem and the difficulty of estimating the density could be reduced if the mask is known. A key to the proposed scheme is the strategy of incorporating the mask prediction into the density map estimator. To this end, we study five possible solutions, and via analysis and experimental validation we identify the most effective one. Through extensive experiments on five public datasets, we demonstrate the superior performance of the proposed approach over the baselines and show that our network could achieve the state-of-the-art performance.
《A Remote Sensing Image Dataset for Cloud Removal》
arXiv：https://arxiv.org/abs/1901.00600
datasets：https://github.com/BUPTLdy/RICE_DATASET
Cloud-based overlays are often present in optical remote sensing images, thus limiting the application of acquired data. Removing clouds is an indispensable pre-processing step in remote sensing image analysis. Deep learning has achieved great success in the field of remote sensing in recent years, including scene classification and change detection. However, deep learning is rarely applied in remote sensing image removal clouds. The reason is the lack of data sets for training neural networks. In order to solve this problem, this paper first proposed the Remote sensing Image Cloud rEmoving dataset (RICE). The proposed dataset consists of two parts: RICE1 contains 500 pairs of images, each pair has images with cloud and cloudless size of 512512; RICE2 contains 450 sets of images, each set contains three 512512 size images. , respectively, the reference picture without clouds, the picture of the cloud and the mask of its cloud. The dataset is freely available at https://github.com/BUPTLdy/RICE_DATASET.
《Fast Perceptual Image Enhancement》
arXiv：https://arxiv.org/abs/1812.11852
The vast majority of photos taken today are by mobile phones. While their quality is rapidly growing, due to physical limitations and cost constraints, mobile phone cameras struggle to compare in quality with DSLR cameras. This motivates us to computationally enhance these images. We extend upon the results of Ignatov et al., where they are able to translate images from compact mobile cameras into images with comparable quality to high-resolution photos taken by DSLR cameras. However, the neural models employed require large amounts of computational resources and are not lightweight enough to run on mobile devices. We build upon the prior work and explore different network architectures targeting an increase in image quality and speed. With an efficient network architecture which does most of its processing in a lower spatial resolution, we achieve a significantly higher mean opinion score (MOS) than the baseline while speeding up the computation by 6.3 times on a consumer-grade CPU. This suggests a promising direction for neural-network-based photo enhancement using the phone hardware of the future.
《Actor Conditioned Attention Maps for Video Action Detection
arXiv：https://arxiv.org/abs/1812.11631
Interactions with surrounding objects and people contain important information towards understanding human actions. In order to model such interactions explicitly, we propose to generate attention maps that rank each spatio-temporal region's importance to a detected actor. We refer to these as Actor-Conditioned Attention Maps (ACAM), and these maps serve as weights to the features extracted from the whole scene. These resulting actor-conditioned features help focus the learned model on regions that are important/relevant to the conditioned actor. Another novelty of our approach is in the use of pre-trained object detectors, instead of region proposals, that generalize better to videos from different sources. Detailed experimental results on the AVA 2.1 datasets demonstrate the importance of interactions, with a performance improvement of 5 mAP with respect to state of the art published results.
《BNN+: Improved Binary Network Training》
arXiv：https://arxiv.org/abs/1812.11800
Deep neural networks (DNN) are widely used in many applications. However, their deployment on edge devices has been difficult because they are resource hungry. Binary neural networks (BNN) help to alleviate the prohibitive resource requirements of DNN, where both activations and weights are limited to 1-bit. We propose an improved binary training method (BNN+), by introducing a regularization function that encourages training weights around binary values. In addition to this, to enhance model performance we add trainable scaling factors to our regularization functions. Furthermore, we use an improved approximation of the derivative of the sign activation function in the backward computation. These additions are based on linear operations that are easily implementable into the binary training framework. We show experimental results on CIFAR-10 obtaining an accuracy of 86.7%, on AlexNet and 91.3% with VGG network. On ImageNet, our method also outperforms the traditional BNN method and XNOR-net, using AlexNet by a margin of 4% and 2% top-1 accuracy respectively.
《Kymatio: Scattering Transforms in Python》
arXiv：https://arxiv.org/abs/1812.11214
homepage：https://www.kymat.io/
The wavelet scattering transform is an invariant signal representation suitable for many signal processing and machine learning applications. We present the Kymatio software package, an easy-to-use, high-performance Python implementation of the scattering transform in 1D, 2D, and 3D that is compatible with modern deep learning frameworks. All transforms may be executed on a GPU (in addition to CPU), offering a considerable speed up over CPU implementations. The package also has a small memory footprint, resulting inefficient memory usage. The source code, documentation, and examples are available undera BSD license at this https URL
《An introduction to domain adaptation and transfer learning》
arXiv：https://arxiv.org/abs/1812.11806
In machine learning, if the training data is an unbiased sample of an underlying distribution, then the learned classification function will make accurate predictions for new samples. However, if the training data is not an unbiased sample, then there will be differences between how the training data is distributed and how the test data is distributed. Standard classifiers cannot cope with changes in data distributions between training and test phases, and will not perform well. Domain adaptation and transfer learning are sub-fields within machine learning that are concerned with accounting for these types of changes. Here, I present an introduction to these fields, guided by the question: when and how can a classifier generalize from a source to a target domain? I will start with a brief introduction into risk minimization, and how transfer learning and domain adaptation expand upon this framework. Following that, I discuss three special cases of data set shift, namely prior, covariate and concept shift. For more complex domain shifts, there are a wide variety of approaches. These are categorized into: importance-weighting, subspace mapping, domain-invariant spaces, feature augmentation, minimax estimators and robust algorithms. A number of points will arise, which I will discuss in the last section. I conclude with the remark that many open questions will have to be addressed before transfer learners and domain-adaptive classifiers become practical.
《Learning Efficient Detector with Semi-supervised Adaptive Distillation》
arXiv：https://arxiv.org/abs/1901.00366
github：https://github.com/Tangshitao/Semi-supervised-Adaptive-Distillation
Knowledge Distillation (KD) has been used in image classification for model compression. However, rare studies apply this technology on single-stage object detectors. Focal loss shows that the accumulated errors of easily-classified samples dominate the overall loss in the training process. This problem is also encountered when applying KD in the detection task. For KD, the teacher-defined hard samples are far more important than any others. We propose ADL to address this issue by adaptively mimicking the teacher's logits, with more attention paid on two types of hard samples: hard-to-learn samples predicted by teacher with low certainty and hard-to-mimic samples with a large gap between the teacher's and the student's prediction. ADL enlarges the distillation loss for hard-to-learn and hard-to-mimic samples and reduces distillation loss for the dominant easy samples, enabling distillation to work on the single-stage detector first time, even if the student and the teacher are identical. Besides, ADL is effective in both the supervised setting and the semi-supervised setting, even when the labeled data and unlabeled data are from different distributions. For distillation on unlabeled data, ADL achieves better performance than existing data distillation which simply utilizes hard targets, making the student detector surpass its teacher. On the COCO database, semi-supervised adaptive distillation (SAD) makes a student detector with a backbone of ResNet-50 surpasses its teacher with a backbone of ResNet-101, while the student has half of the teacher's computation complexity. The code is avaiable at https://github.com/Tangshitao/Semi-supervised-Adaptive-Distillation
《On Minimum Discrepancy Estimation for Deep Domain Adaptation》
Accepted in Joint IJCAI/ECAI/AAMAS/ICML 2018 Workshop
arXiv：https://arxiv.org/abs/1901.00282
In the presence of large sets of labeled data, Deep Learning (DL) has accomplished extraordinary triumphs in the avenue of computer vision, particularly in object classification and recognition tasks. However, DL cannot always perform well when the training and testing images come from different distributions or in the presence of domain shift between training and testing images. They also suffer in the absence of labeled input data. Domain adaptation (DA) methods have been proposed to make up the poor performance due to domain shift. In this paper, we present a new unsupervised deep domain adaptation method based on the alignment of second order statistics (covariances) as well as maximum mean discrepancy of the source and target data with a two stream Convolutional Neural Network (CNN). We demonstrate the ability of the proposed approach to achieve state-of the-art performance for image classification on three benchmark domain adaptation datasets: Office-31 [27], Office-Home [37] and Office-Caltech [8].
《EdgeConnect: Generative Image Inpainting with Adversarial Edge Learning》
arXiv：https://arxiv.org/abs/1901.00212
github：https://arxiv.org/abs/1901.00212
Over the last few years, deep learning techniques have yielded significant improvements in image inpainting. However, many of these techniques fail to reconstruct reasonable structures as they are commonly over-smoothed and/or blurry. This paper develops a new approach for image inpainting that does a better job of reproducing filled regions exhibiting fine details. We propose a two-stage adversarial model EdgeConnect that comprises of an edge generator followed by an image completion network. The edge generator hallucinates edges of the missing region (both regular and irregular) of the image, and the image completion network fills in the missing regions using hallucinated edges as a priori. We evaluate our model end-to-end over the publicly available datasets CelebA, Places2, and Paris StreetView, and show that it outperforms current state-of-the-art techniques quantitatively and qualitatively.
《Not All Words are Equal: Video-specific Information Loss for Video Captioning》
arXiv：https://arxiv.org/abs/1901.00097
An ideal description for a given video should fix its gaze on salient and representative content, which is capable of distinguishing this video from others. However, the distribution of different words is unbalanced in video captioning datasets, where distinctive words for describing video-specific salient objects are far less than common words such as 'a' 'the' and 'person'. The dataset bias often results in recognition error or detail deficiency of salient but unusual objects. To address this issue, we propose a novel learning strategy called Information Loss, which focuses on the relationship between the video-specific visual content and corresponding representative words. Moreover, a framework with hierarchical visual representations and an optimized hierarchical attention mechanism is established to capture the most salient spatial-temporal visual information, which fully exploits the potential strength of the proposed learning strategy. Extensive experiments demonstrate that the ingenious guidance strategy together with the optimized architecture outperforms state-of-the-art video captioning methods on MSVD with CIDEr score 87.5, and achieves superior CIDEr score 47.7 on MSR-VTT. We also show that our Information Loss is generic which improves various models by significant margins.
《Extreme Relative Pose Estimation for RGB-D Scans via Scene Completion》
arXiv：https://arxiv.org/abs/1901.00063
Estimating the relative rigid pose between two RGB-D scans of the same underlying environment is a fundamental problem in computer vision, robotics, and computer graphics. Most existing approaches allow only limited maximum relative pose changes since they require considerable overlap between the input scans. We introduce a novel deep neural network that extends the scope to extreme relative poses, with little or even no overlap between the input scans. The key idea is to infer more complete scene information about the underlying environment and match on the completed scans. In particular, instead of only performing scene completion from each individual scan, our approach alternates between relative pose estimation and scene completion. This allows us to perform scene completion by utilizing information from both input scans at late iterations, resulting in better results for both scene completion and relative pose estimation. Experimental results on benchmark datasets show that our approach leads to considerable improvements over state-of-the-art approaches for relative pose estimation. In particular, our approach provides encouraging relative pose estimates even between non-overlapping scans.
《A Survey on Multi-output Learning》
arXiv：https://arxiv.org/abs/1901.00248
Multi-output learning aims to simultaneously predict multiple outputs given an input. It is an important learning problem due to the pressing need for sophisticated decision making in real-world applications. Inspired by big data, the 4Vs characteristics of multi-output imposes a set of challenges to multi-output learning, in terms of the volume, velocity, variety and veracity of the outputs. Increasing number of works in the literature have been devoted to the study of multi-output learning and the development of novel approaches for addressing the challenges encountered. However, it lacks a comprehensive overview on different types of challenges of multi-output learning brought by the characteristics of the multiple outputs and the techniques proposed to overcome the challenges. This paper thus attempts to fill in this gap to provide a comprehensive review on this area. We first introduce different stages of the life cycle of the output labels. Then we present the paradigm on multi-output learning, including its myriads of output structures, definitions of its different sub-problems, model evaluation metrics and popular data repositories used in the study. Subsequently, we review a number of state-of-the-art multi-output learning methods, which are categorized based on the challenges.
《Deep Frame Prediction for Video Coding》
submitted to IEEE Trans
arXiv：https://arxiv.org/abs/1901.00062
We propose a novel frame prediction method using a deep neural network (DNN), with the goal of improving video coding efficiency. The proposed DNN makes use of decoded frames, at both encoder and decoder, to predict textures of the current coding block. Unlike conventional inter-prediction, the proposed method does not require any motion information to be transferred between the encoder and the decoder. Still, both uni-directional and bi-directional prediction are possible using the proposed DNN, which is enabled by the use of the temporal index channel, in addition to color channels. In this study, we developed a jointly trained DNN for both uni- and bi-directional prediction, as well as separate networks for uni- and bi-directional prediction, and compared the efficacy of both approaches. The proposed DNNs were compared with the conventional motion-compensated prediction in the latest video coding standard, HEVC, in terms of BD-Bitrate. The experiments show that the proposed joint DNN (for both uni- and bi-directional prediction) reduces the luminance bitrate by about 3.9%, 2.2%, and 2.2% in the Low delay P, Low delay, and Random access configurations, respectively. In addition, using the separately trained DNNs brings further bit savings of about 0.4%-0.8%.
《CLEVR-Ref+: Diagnosing Visual Reasoning with Referring Expressions》
arXiv：https://arxiv.org/abs/1901.00850
github & datasets：https://cs.jhu.edu/~cxliu/2019/clevr-ref+
Referring object detection and referring image segmentation are important tasks that require joint understanding of visual information and natural language. Yet there has been evidence that current benchmark datasets suffer from bias, and current state-of-the-art models cannot be easily evaluated on their intermediate reasoning process. To address these issues and complement similar efforts in visual question answering, we build CLEVR-Ref+, a synthetic diagnostic dataset for referring expression comprehension. The precise locations and attributes of the objects are readily available, and the referring expressions are automatically associated with functional programs. The synthetic nature allows control over dataset bias (through sampling strategy), and the modular programs enable intermediate reasoning ground truth without human annotators. In addition to evaluating several state-of-the-art models on CLEVR-Ref+, we also propose IEP-Ref, a module network approach that significantly outperforms other models on our dataset. In particular, we present two interesting and important findings using IEP-Ref: (1) the module trained to transform feature maps into segmentation masks can be attached to any intermediate module to reveal the entire reasoning process step-by-step; (2) even if all training data has at least one object referred, IEP-Ref can correctly predict no-foreground when presented with false-premise referring expressions. To the best of our knowledge, this is the first direct and quantitative proof that neural modules behave in the way they are intended.
《A Hierarchical Grocery Store Image Dataset with Visual and Semantic Labels》
WACV 2019
arXiv：https://arxiv.org/abs/1901.00711
Image classification models built into visual support systems and other assistive devices need to provide accurate predictions about their environment. We focus on an application of assistive technology for people with visual impairments, for daily activities such as shopping or cooking. In this paper, we provide a new benchmark dataset for a challenging task in this application - classification of fruits, vegetables, and refrigerated products, e.g. milk packages and juice cartons, in grocery stores. To enable the learning process to utilize multiple sources of structured information, this dataset not only contains a large volume of natural images but also includes the corresponding information of the product from an online shopping website. Such information encompasses the hierarchical structure of the object classes, as well as an iconic image of each type of object. This dataset can be used to train and evaluate image classification models for helping visually impaired people in natural environments. Additionally, we provide benchmark results evaluated on pretrained convolutional neural networks often used for image understanding purposes, and also a multi-view variational autoencoder, which is capable of utilizing the rich product information in the dataset.
《Active Learning with TensorBoard Projector》
arXiv：https://arxiv.org/abs/1901.00675
An ML-based system for interactive labeling of image datasets is contributed in TensorBoard Projector to speed up image annotation performed by humans. The tool visualizes feature spaces and makes it directly editable by online integration of applied labels, and it is a system for verifying and managing machine learning data pertaining to labels. We propose realistic annotation emulation to evaluate the system design of interactive active learning, based on our improved semi-supervised extension of t-SNE dimensionality reduction. Our active learning tool can significantly increase labeling efficiency compared to uncertainty sampling, and we show that less than 100 labeling actions are typically sufficient for good classification on a variety of specialized image datasets. Our contribution is unique given that it needs to perform dimensionality reduction, feature space visualization and editing, interactive label propagation, low-complexity active learning, human perceptual modeling, annotation emulation and unsupervised feature extraction for specialized datasets in a production-quality implementation.
《Edge-Semantic Learning Strategy for Layout Estimation in Indoor Environment》
arXiv：https://arxiv.org/abs/1901.00621
Visual cognition of the indoor environment can benefit from the spatial layout estimation, which is to represent an indoor scene with a 2D box on a monocular image. In this paper, we propose to fully exploit the edge and semantic information of a room image for layout estimation. More specifically, we present an encoder-decoder network with shared encoder and two separate decoders, which are composed of multiple deconvolution (transposed convolution) layers, to jointly learn the edge maps and semantic labels of a room image. We combine these two network predictions in a scoring function to evaluate the quality of the layouts, which are generated by ray sampling and from a predefined layout pool. Guided by the scoring function, we apply a novel refinement strategy to further optimize the layout hypotheses. Experimental results show that the proposed network can yield accurate estimates of edge maps and semantic labels. By fully utilizing the two different types of labels, the proposed method achieves state-of-the-art layout estimation performance on benchmark datasets.
《Photo-Sketching: Inferring Contour Drawings from Images》
homepage(include code)：http:https://www.cs.cmu.edu/~mengtial/proj/sketch/
WACV 2019
arXiv：https://arxiv.org/abs/1901.00542
Edges, boundaries and contours are important subjects of study in both computer graphics and computer vision. On one hand, they are the 2D elements that convey 3D shapes, on the other hand, they are indicative of occlusion events and thus separation of objects or semantic concepts. In this paper, we aim to generate contour drawings, boundary-like drawings that capture the outline of the visual scene. Prior art often cast this problem as boundary detection. However, the set of visual cues presented in the boundary detection output are different from the ones in contour drawings, and also the artistic style is ignored. We address these issues by collecting a new dataset of contour drawings and proposing a learning-based method that resolves diversity in the annotation and, unlike boundary detectors, can work with imperfect alignment of the annotation and the actual ground truth. Our method surpasses previous methods quantitatively and qualitatively. Surprisingly, when our model fine-tunes on BSDS500, we achieve the state-of-the-art performance in salient boundary detection, suggesting contour drawing might be a scalable alternative to boundary annotation, which at the same time is easier and more interesting for annotators to draw.
《Visualizing Deep Similarity Networks》
arXiv：https://arxiv.org/abs/1901.00536
For convolutional neural network models that optimize an image embedding, we propose a method to highlight the regions of images that contribute most to pairwise similarity. This work is a corollary to the visualization tools developed for classification networks, but applicable to the problem domains better suited to similarity learning. The visualization shows how similarity networks that are fine-tuned learn to focus on different features. We also generalize our approach to embedding networks that use different pooling strategies and provide a simple mechanism to support image similarity searches on objects or sub-regions in the query image.
《Resource-Scalable CNN Synthesis for IoT Applications》
arXiv：https://arxiv.org/abs/1901.00738
State-of-the-art image recognition systems use sophisticated Convolutional Neural Networks (CNNs) that are designed and trained to identify numerous object classes. Such networks are fairly resource intensive to compute, prohibiting their deployment on resource-constrained embedded platforms. On one hand, the ability to classify an exhaustive list of categories is excessive for the demands of most IoT applications. On the other hand, designing a new custom-designed CNN for each new IoT application is impractical, due to the inherent difficulty in developing competitive models and time-to-market pressure. To address this problem, we investigate the question of: "Can one utilize an existing optimized CNN model to automatically build a competitive CNN for an IoT application whose objects of interest are a fraction of categories that the original CNN was designed to classify, such that the resource requirement is proportionally scaled down?" We use the term resource scalability to refer to this concept, and develop a methodology for automated synthesis of resource scalable CNNs from an existing optimized baseline CNN. The synthesized CNN has sufficient learning capacity for handling the given IoT application requirements, and yields competitive accuracy. The proposed approach is fast, and unlike the presently common practice of CNN design, does not require iterative rounds of training trial and error.
《Linear colour segmentation revisited》
注：颜色分割，有意思的研究方向
arXiv：https://arxiv.org/abs/1901.00534
github：https://github.com/visillect/segmentation
In this work we discuss the known algorithms for linear colour segmentation based on a physical approach and propose a new modification of segmentation algorithm. This algorithm is based on a region adjacency graph framework without a pre-segmentation stage. Proposed edge weight functions are defined from linear image model with normal noise. The colour space projective transform is introduced as a novel pre-processing technique for better handling of shadow and highlight areas. The resulting algorithm is tested on a benchmark dataset consisting of the images of 19 natural scenes selected from the Barnard's DXC-930 SFU dataset and 12 natural scene images newly published for common use. The dataset is provided with pixel-by-pixel ground truth colour segmentation for every image. Using this dataset, we show that the proposed algorithm modifications lead to qualitative advantages over other model-based segmentation algorithms, and also show the positive effect of each proposed modification. The source code and datasets for this work are available for free access at https://github.com/visillect/segmentation.
《Visualizing Deep Similarity Networks》
arXiv：https://arxiv.org/abs/1901.00536
For convolutional neural network models that optimize an image embedding, we propose a method to highlight the regions of images that contribute most to pairwise similarity. This work is a corollary to the visualization tools developed for classification networks, but applicable to the problem domains better suited to similarity learning. The visualization shows how similarity networks that are fine-tuned learn to focus on different features. We also generalize our approach to embedding networks that use different pooling strategies and provide a simple mechanism to support image similarity searches on objects or sub-regions in the query image.