Vision is the richest and most cost-effective technology for Driver Monitoring Systems (DMS), esp... more Vision is the richest and most cost-effective technology for Driver Monitoring Systems (DMS), especially after the recent success of Deep Learning (DL) methods. The lack of sufficiently large and comprehensive datasets is currently a bottleneck for the progress of DMS development, crucial for the transition of automated driving from SAE Level-2 to SAE Level-3. In this paper, we introduce the Driver Monitoring Dataset (DMD), an extensive dataset which includes real and simulated driving scenarios: distraction, gaze allocation, drowsiness, hands-wheel interaction and context data, in 41 hours of RGB, depth and IR videos from 3 cameras capturing face, body and hands of 37 drivers. A comparison with existing similar datasets is included, which shows the DMD is more extensive, diverse, and multipurpose. The usage of the DMD is illustrated by extracting a subset of it, the dBehaviourMD dataset, containing 13 distraction activities, prepared to be used in DL training processes. Furthermore, we propose a robust and real-time driver behaviour recognition system targeting a real-world application that can run on cost-efficient CPU-only platforms, based on the dBehaviourMD. Its performance is evaluated with different types of fusion strategies, which all reach enhanced accuracy still providing real-time response.
2010 IEEE Workshop On Signal Processing Systems, 2010
Particle filtering is widely used in numerous nonlinear applications which require reconfigurabil... more Particle filtering is widely used in numerous nonlinear applications which require reconfigurability, fast prototyping, and online parallel signal processing. The emerging computing platform, CUDA, may be regarded as the most appealing platform for such implementation. However, there are not yet literatures exploring how to utilize CUDA for particle filters. This parer aims to provide two design techniques, A) finiteredraw importance-maximizing (FRIM) prior editing and B) localized resampling, for efficient implementation of particle filters on CUDA, which can be verified to reduce global operations and provide significant speedup. The modifications on algorithm and architectural mapping are evaluated with conceptual and quantitative analysis. From the classic bearingsonly tracking experiments, the proposed design is 5.73 times faster than the direct implementation on GeForce 9400m.
Many recent reconfigurable/multi-mode quasi-cyclic low density parity check (QC-LDPC) decoder des... more Many recent reconfigurable/multi-mode quasi-cyclic low density parity check (QC-LDPC) decoder designs have shown appealing implementation results in the literature. However, most of them are based on datapath multiplexing techniques with ad hoc matrix arrangement. There is still room for further interconnection reduction, throughput enhancement, and a more sophisticated early termination scheme. In this paper, we will focus on these issues and present a two-level design approach, which optimizes the design at (1) matrix merging level, and (2) module design level. First, direct multiplexing datapaths between multiple modes leads to great overhead on wiring complexity. In order to mitigate this problem, we merge multiple parity check matrices by proposing an efficient algorithm at matrix merging level, which helps to minimize multiplexer and wiring overhead. Second, for efficient decoding issues, we propose two design techniques at module design level. One is data wrapping scheme. It ...
2010 IEEE Workshop On Signal Processing Systems, 2010
Constrained particle filter is widely used in indoor localization applications. With environmenta... more Constrained particle filter is widely used in indoor localization applications. With environmental information, the cascaded hypothesis modifier can improve accuracy by rejecting particles those have invalid transitions. However, the memory requirement and computation complexity of constrained particle filter are both large, and the low spatial correlations between the sequentially accessed particles make the computation inefficient. This paper proposes two techniques to improve the efficiency: location constrained multi-prediction and dynamically cascaded hypothesis modifier. The experimental result shows that the proposed techniques can achieve higher accuracy at lower cost, both in storage and computation.
Vision is the richest and most cost-effective technology for Driver Monitoring Systems (DMS), esp... more Vision is the richest and most cost-effective technology for Driver Monitoring Systems (DMS), especially after the recent success of Deep Learning (DL) methods. The lack of sufficiently large and comprehensive datasets is currently a bottleneck for the progress of DMS development, crucial for the transition of automated driving from SAE Level-2 to SAE Level-3. In this paper, we introduce the Driver Monitoring Dataset (DMD), an extensive dataset which includes real and simulated driving scenarios: distraction, gaze allocation, drowsiness, hands-wheel interaction and context data, in 41 hours of RGB, depth and IR videos from 3 cameras capturing face, body and hands of 37 drivers. A comparison with existing similar datasets is included, which shows the DMD is more extensive, diverse, and multi-purpose. The usage of the DMD is illustrated by extracting a subset of it, the dBehaviourMD dataset, containing 13 distraction activities, prepared to be used in DL training processes. Furthermor...
This paper shows a triple-mode LDPC decoder design with two design techniques, the matrix reorder... more This paper shows a triple-mode LDPC decoder design with two design techniques, the matrix reordering algorithm for multi-mode reconfiguration and the Single-Entry-Multiple-Data (SEMD) scheme for throughput enhancement. The matrix reordering algorithm can reduce the computational complexity from O(n!) to O(n3). The SEMD can enhance the throughput by m times with small area overhead. With TSMC 0.13µm CMOS, the proposed design is synthesized in 1.99mm2 area at 172.4MHz.
Particle filtering is widely used in numerous nonlinear applications which require reconfigurabil... more Particle filtering is widely used in numerous nonlinear applications which require reconfigurability, fast prototyping, and online parallel signal processing. The emerging computing platform, CUDA, may be regarded as the most appealing platform for such implementation. However, there are not yet literatures exploring how to utilize CUDA for particle filters. This parer aims to provide two design techniques, A) finite-redraw importance-maximizing
Eurasip Journal on Advances in Signal Processing, 2011
Particle filter (PF) is an emerging signal processing methodology, which can effectively deal wit... more Particle filter (PF) is an emerging signal processing methodology, which can effectively deal with nonlinear and non-Gaussian signals by a sample-based approximation of the state probability density function. The particle generation of the PF is a data-independent procedure and can be implemented in parallel. However, the resampling procedure in the PF is a sequential task in natural and difficult to be parallelized. Based on the Amdahl's law, the sequential portion of a task limits the maximum speed-up of the parallelized implementation. Moreover, large particle number is usually required to obtain an accurate estimation, and the complexity of the resampling procedure is highly related to the number of particles. In this article, we propose a multi-prediction (MP) framework with two selection approaches. The proposed MP framework can reduce the required particle number for target estimation accuracy, and the sequential operation of the resampling can be reduced. Besides, the overhead of the MP framework can be easily compensated by parallel implementation. The proposed MP-PF alleviates the global sequential operation by increasing the local parallel computation. In addition, the MP-PF is very suitable for multi-core graphics processing unit (GPU) platform, which is a popular parallel processing architecture. We give prototypical implementations of the MP-PFs on multi-core GPU platform. For the classic bearing-only tracking experiments, the proposed MP-PF can be 25.1 and 15.3 times faster than the sequential importance resampling-PF with 10,000 and 20,000 particles, respectively. Hence, the proposed MP-PF can enhance the efficiency of the parallelization.
Vision is the richest and most cost-effective technology for Driver Monitoring Systems (DMS), esp... more Vision is the richest and most cost-effective technology for Driver Monitoring Systems (DMS), especially after the recent success of Deep Learning (DL) methods. The lack of sufficiently large and comprehensive datasets is currently a bottleneck for the progress of DMS development, crucial for the transition of automated driving from SAE Level-2 to SAE Level-3. In this paper, we introduce the Driver Monitoring Dataset (DMD), an extensive dataset which includes real and simulated driving scenarios: distraction, gaze allocation, drowsiness, hands-wheel interaction and context data, in 41 hours of RGB, depth and IR videos from 3 cameras capturing face, body and hands of 37 drivers. A comparison with existing similar datasets is included, which shows the DMD is more extensive, diverse, and multi-purpose. The usage of the DMD is illustrated by extracting a subset of it, the dBehaviourMD dataset, containing 13 distraction activities, prepared to be used in DL training processes. Furthermor...
Vision is the richest and most cost-effective technology for Driver Monitoring Systems (DMS), esp... more Vision is the richest and most cost-effective technology for Driver Monitoring Systems (DMS), especially after the recent success of Deep Learning (DL) methods. The lack of sufficiently large and comprehensive datasets is currently a bottleneck for the progress of DMS development, crucial for the transition of automated driving from SAE Level-2 to SAE Level-3. In this paper, we introduce the Driver Monitoring Dataset (DMD), an extensive dataset which includes real and simulated driving scenarios: distraction, gaze allocation, drowsiness, hands-wheel interaction and context data, in 41 hours of RGB, depth and IR videos from 3 cameras capturing face, body and hands of 37 drivers. A comparison with existing similar datasets is included, which shows the DMD is more extensive, diverse, and multipurpose. The usage of the DMD is illustrated by extracting a subset of it, the dBehaviourMD dataset, containing 13 distraction activities, prepared to be used in DL training processes. Furthermore, we propose a robust and real-time driver behaviour recognition system targeting a real-world application that can run on cost-efficient CPU-only platforms, based on the dBehaviourMD. Its performance is evaluated with different types of fusion strategies, which all reach enhanced accuracy still providing real-time response.
2010 IEEE Workshop On Signal Processing Systems, 2010
Particle filtering is widely used in numerous nonlinear applications which require reconfigurabil... more Particle filtering is widely used in numerous nonlinear applications which require reconfigurability, fast prototyping, and online parallel signal processing. The emerging computing platform, CUDA, may be regarded as the most appealing platform for such implementation. However, there are not yet literatures exploring how to utilize CUDA for particle filters. This parer aims to provide two design techniques, A) finiteredraw importance-maximizing (FRIM) prior editing and B) localized resampling, for efficient implementation of particle filters on CUDA, which can be verified to reduce global operations and provide significant speedup. The modifications on algorithm and architectural mapping are evaluated with conceptual and quantitative analysis. From the classic bearingsonly tracking experiments, the proposed design is 5.73 times faster than the direct implementation on GeForce 9400m.
Many recent reconfigurable/multi-mode quasi-cyclic low density parity check (QC-LDPC) decoder des... more Many recent reconfigurable/multi-mode quasi-cyclic low density parity check (QC-LDPC) decoder designs have shown appealing implementation results in the literature. However, most of them are based on datapath multiplexing techniques with ad hoc matrix arrangement. There is still room for further interconnection reduction, throughput enhancement, and a more sophisticated early termination scheme. In this paper, we will focus on these issues and present a two-level design approach, which optimizes the design at (1) matrix merging level, and (2) module design level. First, direct multiplexing datapaths between multiple modes leads to great overhead on wiring complexity. In order to mitigate this problem, we merge multiple parity check matrices by proposing an efficient algorithm at matrix merging level, which helps to minimize multiplexer and wiring overhead. Second, for efficient decoding issues, we propose two design techniques at module design level. One is data wrapping scheme. It ...
2010 IEEE Workshop On Signal Processing Systems, 2010
Constrained particle filter is widely used in indoor localization applications. With environmenta... more Constrained particle filter is widely used in indoor localization applications. With environmental information, the cascaded hypothesis modifier can improve accuracy by rejecting particles those have invalid transitions. However, the memory requirement and computation complexity of constrained particle filter are both large, and the low spatial correlations between the sequentially accessed particles make the computation inefficient. This paper proposes two techniques to improve the efficiency: location constrained multi-prediction and dynamically cascaded hypothesis modifier. The experimental result shows that the proposed techniques can achieve higher accuracy at lower cost, both in storage and computation.
Vision is the richest and most cost-effective technology for Driver Monitoring Systems (DMS), esp... more Vision is the richest and most cost-effective technology for Driver Monitoring Systems (DMS), especially after the recent success of Deep Learning (DL) methods. The lack of sufficiently large and comprehensive datasets is currently a bottleneck for the progress of DMS development, crucial for the transition of automated driving from SAE Level-2 to SAE Level-3. In this paper, we introduce the Driver Monitoring Dataset (DMD), an extensive dataset which includes real and simulated driving scenarios: distraction, gaze allocation, drowsiness, hands-wheel interaction and context data, in 41 hours of RGB, depth and IR videos from 3 cameras capturing face, body and hands of 37 drivers. A comparison with existing similar datasets is included, which shows the DMD is more extensive, diverse, and multi-purpose. The usage of the DMD is illustrated by extracting a subset of it, the dBehaviourMD dataset, containing 13 distraction activities, prepared to be used in DL training processes. Furthermor...
This paper shows a triple-mode LDPC decoder design with two design techniques, the matrix reorder... more This paper shows a triple-mode LDPC decoder design with two design techniques, the matrix reordering algorithm for multi-mode reconfiguration and the Single-Entry-Multiple-Data (SEMD) scheme for throughput enhancement. The matrix reordering algorithm can reduce the computational complexity from O(n!) to O(n3). The SEMD can enhance the throughput by m times with small area overhead. With TSMC 0.13µm CMOS, the proposed design is synthesized in 1.99mm2 area at 172.4MHz.
Particle filtering is widely used in numerous nonlinear applications which require reconfigurabil... more Particle filtering is widely used in numerous nonlinear applications which require reconfigurability, fast prototyping, and online parallel signal processing. The emerging computing platform, CUDA, may be regarded as the most appealing platform for such implementation. However, there are not yet literatures exploring how to utilize CUDA for particle filters. This parer aims to provide two design techniques, A) finite-redraw importance-maximizing
Eurasip Journal on Advances in Signal Processing, 2011
Particle filter (PF) is an emerging signal processing methodology, which can effectively deal wit... more Particle filter (PF) is an emerging signal processing methodology, which can effectively deal with nonlinear and non-Gaussian signals by a sample-based approximation of the state probability density function. The particle generation of the PF is a data-independent procedure and can be implemented in parallel. However, the resampling procedure in the PF is a sequential task in natural and difficult to be parallelized. Based on the Amdahl's law, the sequential portion of a task limits the maximum speed-up of the parallelized implementation. Moreover, large particle number is usually required to obtain an accurate estimation, and the complexity of the resampling procedure is highly related to the number of particles. In this article, we propose a multi-prediction (MP) framework with two selection approaches. The proposed MP framework can reduce the required particle number for target estimation accuracy, and the sequential operation of the resampling can be reduced. Besides, the overhead of the MP framework can be easily compensated by parallel implementation. The proposed MP-PF alleviates the global sequential operation by increasing the local parallel computation. In addition, the MP-PF is very suitable for multi-core graphics processing unit (GPU) platform, which is a popular parallel processing architecture. We give prototypical implementations of the MP-PFs on multi-core GPU platform. For the classic bearing-only tracking experiments, the proposed MP-PF can be 25.1 and 15.3 times faster than the sequential importance resampling-PF with 10,000 and 20,000 particles, respectively. Hence, the proposed MP-PF can enhance the efficiency of the parallelization.
Vision is the richest and most cost-effective technology for Driver Monitoring Systems (DMS), esp... more Vision is the richest and most cost-effective technology for Driver Monitoring Systems (DMS), especially after the recent success of Deep Learning (DL) methods. The lack of sufficiently large and comprehensive datasets is currently a bottleneck for the progress of DMS development, crucial for the transition of automated driving from SAE Level-2 to SAE Level-3. In this paper, we introduce the Driver Monitoring Dataset (DMD), an extensive dataset which includes real and simulated driving scenarios: distraction, gaze allocation, drowsiness, hands-wheel interaction and context data, in 41 hours of RGB, depth and IR videos from 3 cameras capturing face, body and hands of 37 drivers. A comparison with existing similar datasets is included, which shows the DMD is more extensive, diverse, and multi-purpose. The usage of the DMD is illustrated by extracting a subset of it, the dBehaviourMD dataset, containing 13 distraction activities, prepared to be used in DL training processes. Furthermor...
Uploads
Papers by Min-An Chao