YOLOv4-dense: A smaller and faster YOLOv4 for real-time edge-device based object detection in traffic scene
文献类型:期刊论文
作者 | Jiang, Yue1,2,3; Li, Wenjing2,3; Zhang, Jun2,3; Li, Fang2,3; Wu, Zhongcheng2,3 |
刊名 | IET IMAGE PROCESSING |
出版日期 | 2022-11-08 |
ISSN号 | 1751-9659 |
DOI | 10.1049/ipr2.12656 |
通讯作者 | Zhang, Jun(zhang_jun@hmfl.ac.cn) |
英文摘要 | Edge-device-based object detection is crucial in many real-world applications, such as self-driving cars, ADAS, driver behavior analysis. Although deep learning (DL) has become the de-facto approach for object detection, the limited computing resources of embedded devices and the large model size of current DL-based methods increase the difficulty of real-time object detection on edge devices. To overcome these difficulties, in this work a novel YOLOv4-dense model is proposed to detect objects in an accurate, fast manner, which is built on top of the YOLOv4 framework but with substantial improvements. More specifically, lots of CSP layers are pruned since it will decrease inference speed. And to address the losing small objects problem, a dense block is introduced. In addition, a lightweight two-stream YOLO head is also designed to further reduce the computational complexity of the model. Experimental results on NVIDIA JETSON TX2 embedded platform demonstrate that YOLOv4-dense can achieve a higher accuracy, faster speed with smaller model size. For instance, on the KITTI dataset, YOLOv4-dense obtains 84.3% mAP and 22.6 FPS with only 20.3 M parameters, surpassing the state-of-the-art models with comparable parameter budget such as YOLOv3-tiny, YOLOv4-tiny, PP-YOLO-tiny by a large margin. |
资助项目 | Research on Scientific Data Management Method and Key Technology of Large-scale Scientific Facility ; Pre-research Project on Key Technologies of Integrated Experimental Facilities of Steady High Magnetic Field and Optical Spectroscopy ; High Magnetic Field Laboratory of Anhui Province ; High Magnetic Field Laboratory, Chinese Academy of Sciences[2019HSC-KPRD003] |
WOS研究方向 | Computer Science ; Engineering ; Imaging Science & Photographic Technology |
语种 | 英语 |
出版者 | WILEY |
WOS记录号 | WOS:000879795300001 |
资助机构 | Research on Scientific Data Management Method and Key Technology of Large-scale Scientific Facility ; Pre-research Project on Key Technologies of Integrated Experimental Facilities of Steady High Magnetic Field and Optical Spectroscopy ; High Magnetic Field Laboratory of Anhui Province ; High Magnetic Field Laboratory, Chinese Academy of Sciences |
源URL | [http://ir.hfcas.ac.cn:8080/handle/334002/130130] |
专题 | 中国科学院合肥物质科学研究院 |
通讯作者 | Zhang, Jun |
作者单位 | 1.High Magnet Field Lab Anhui Prov, Hefei, Peoples R China 2.Chinese Acad Sci, High Magnet Field Lab, HFIPS, Hefei 230031, Peoples R China 3.Univ Sci & Technol China, Hefei, Peoples R China |
推荐引用方式 GB/T 7714 | Jiang, Yue,Li, Wenjing,Zhang, Jun,et al. YOLOv4-dense: A smaller and faster YOLOv4 for real-time edge-device based object detection in traffic scene[J]. IET IMAGE PROCESSING,2022. |
APA | Jiang, Yue,Li, Wenjing,Zhang, Jun,Li, Fang,&Wu, Zhongcheng.(2022).YOLOv4-dense: A smaller and faster YOLOv4 for real-time edge-device based object detection in traffic scene.IET IMAGE PROCESSING. |
MLA | Jiang, Yue,et al."YOLOv4-dense: A smaller and faster YOLOv4 for real-time edge-device based object detection in traffic scene".IET IMAGE PROCESSING (2022). |
入库方式: OAI收割
来源:合肥物质科学研究院
浏览0
下载0
收藏0
其他版本
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。