feat(detection): add freeanchor hub file (#12)

models/megengine_vision_freeanchor.md

@@ -0,0 +1,102 @@
+---
+template: hub1
+title: FreeAnchor
+summary:
+ en_US: FreeAnchor pre-trained on COCO2017
+ zh_CN: FreeAnchor (COCO2017预训练权重）
+author: MegEngine Team
+tags: [vision, detection]
+github-link: https://github.com/MegEngine/Models/tree/master/official/vision/detection
+---
+
+```python
+from megengine import hub
+model = hub.load(
+ "megengine/models",
+ "freeanchor_res50_coco_1x_800size",
+ pretrained=True,
+ use_cache=False,
+)
+model.eval()
+
+models_api = hub.import_module(
+ "megengine/models",
+ git_host="github.com",
+)
+```
+<!-- section: zh_CN -->
+
+所有预训练模型希望数据被正确预处理。
+模型要求输入BGR的图片, 同时需要等比例缩放到：短边和长边分别不超过800/1333
+最后做归一化处理 (均值为: `[103.530, 116.280, 123.675]`, 标准差为: `[57.375, 57.120, 58.395]`)。
+
+下面是一段处理一张图片的样例代码。
+
+```python
+# Download an example image from the megengine data website
+import urllib
+url, filename = ("https://data.megengine.org.cn/images/cat.jpg", "cat.jpg")
+try: urllib.URLopener().retrieve(url, filename)
+except: urllib.request.urlretrieve(url, filename)
+
+# Read and pre-process the image
+import cv2
+import megengine as mge
+
+image = cv2.imread("cat.jpg")
+data, im_info = models_api.DetEvaluator.process_inputs(image, 800, 1333)
+predictions = model(image=mge.tensor(data), im_info=mge.tensor(im_info))
+print(predictions)
+```
+
+### 模型描述
+
+目前我们提供了在COCO2017数据集上预训练的FreeAnchor模型, 性能如下：
+
+| model | mAP<br>@5-95 |
+| --- | :---: |
+| freeanchor-res50-coco-1x-800size | 38.9 |
+| freeanchor-res101-coco-2x-800size | 43.3 |
+
+### 参考文献
+
+- [FreeAnchor: Learning to Match Anchors for Visual Object Detection](https://arxiv.org/abs/1909.02466) Xiaosong Zhang, Fang Wan, Chang Liu, Rongrong Ji and Qixiang Ye. Neural Information Processing Systems (NeurIPS), 2019. 
+- [Microsoft COCO: Common Objects in Context](https://arxiv.org/abs/1405.0312) Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C Lawrence Zitnick. European Conference on Computer Vision (ECCV), 2014.
+
+<!-- section: en_US -->
+
+All pre-trained models expect input images normalized in the same way,
+i.e. input images must be 3-channel BGR images of shape `(H x W x 3)`, and reszied shortedge/longedge to no more than `800/1333`.
+The images should be normalized using `mean = [103.530, 116.280, 123.675]` and `std = [57.375, 57.120, 58.395])`.
+
+Here's a sample execution.
+
+```python
+# Download an example image from the megengine data website
+import urllib
+url, filename = ("https://data.megengine.org.cn/images/cat.jpg", "cat.jpg")
+try: urllib.URLopener().retrieve(url, filename)
+except: urllib.request.urlretrieve(url, filename)
+
+# Read and pre-process the image
+import cv2
+import megengine as mge
+
+image = cv2.imread("cat.jpg")
+data, im_info = models_api.DetEvaluator.process_inputs(image, 800, 1333)
+predictions = model(image=mge.tensor(data), im_info=mge.tensor(im_info))
+print(predictions)
+```
+
+### Model Description
+
+Currently we provide RetinaNet models pretrained on COCO2017 dataset. The performance can be found in following table.
+
+| model | mAP<br>@5-95 |
+| --- | :---: |
+| freeanchor-res50-coco-1x-800size | 38.9 |
+| freeanchor-res101-coco-2x-800size | 43.3 |
+
+### References
+- [FreeAnchor: Learning to Match Anchors for Visual Object Detection](https://arxiv.org/abs/1909.02466) Xiaosong Zhang, Fang Wan, Chang Liu, Rongrong Ji and Qixiang Ye. Neural Information Processing Systems (NeurIPS), 2019. 
+- [Microsoft COCO: Common Objects in Context](https://arxiv.org/abs/1405.0312) Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C Lawrence Zitnick. European Conference on Computer Vision (ECCV), 2014.