Compact convolutional neural network cascadefor face detection; CEUR Workshop Proceedings; Vol. 1576 : Parallel Computing Technologies 2016, PCT 2016
| Parent link: | CEUR Workshop Proceedings: Online Proceedings for Scientific Conferences and Workshops Vol. 1576 : Parallel Computing Technologies 2016, PCT 2016.— 2016.— [P. 375-387] |
|---|---|
| Hovedforfatter: | |
| Institution som forfatter: | |
| Andre forfattere: | |
| Summary: | Title screen This paper presents a new solution to the frontal face detection problem based on a compact convolutional neural networks cascade. Test results on an FDDB dataset show that it is able to compete with state-of-the-art algorithms. This proposed detector is implemented using three technologies: SSE/AVX/AVX2 instruction sets for Intel CPUs, Nvidia CUDA, and OpenCL. The detection speed of our approach exceeds considerably all the existing CPUbased and GPU-based algorithms. Thanks to its high computational efficiency, our detector can process 4K Ultra HD video stream in real time (up to 27 fps) on mobile platforms while searching objects with a dimension of 60×60 pixels or higher. At the same time, its processing speed is almost independent of the background and the number of objects in a scene. This is achieved by asynchronous computation of stages in the cascade. |
| Sprog: | engelsk |
| Udgivet: |
2016
|
| Fag: | |
| Online adgang: | http://earchive.tpu.ru/handle/11683/36142 http://ceur-ws.org/Vol-1576/150.pdf |
| Format: | Electronisk Book Chapter |
| KOHA link: | https://koha.lib.tpu.ru/cgi-bin/koha/opac-detail.pl?biblionumber=650721 |
| Summary: | Title screen This paper presents a new solution to the frontal face detection problem based on a compact convolutional neural networks cascade. Test results on an FDDB dataset show that it is able to compete with state-of-the-art algorithms. This proposed detector is implemented using three technologies: SSE/AVX/AVX2 instruction sets for Intel CPUs, Nvidia CUDA, and OpenCL. The detection speed of our approach exceeds considerably all the existing CPUbased and GPU-based algorithms. Thanks to its high computational efficiency, our detector can process 4K Ultra HD video stream in real time (up to 27 fps) on mobile platforms while searching objects with a dimension of 60×60 pixels or higher. At the same time, its processing speed is almost independent of the background and the number of objects in a scene. This is achieved by asynchronous computation of stages in the cascade. |
|---|