Name
Abstract | ||
|---|---|---|
To solve the problems of complex model structure, large number of parameters, and high resource consumption that make it difficult to meet the real-time requirements of embedded target detection tasks, this paper proposed a lightweight target detection algorithm based on improved MobileNetv3-YOLOv3. This algorithm uses MobileNetv3 network to replace the backbone of the original YOLOv3 network, and the reduction of network parameters greatly improves the detection speed of the algorithm; the loss function is modified to CIoU to improve the accuracy and detection speed of the network. The experimental results showed that the improved lightweight detection algorithm on the VOC07 + 12 dataset has a 1.55% improvement in mAP and a 2.47 times improvement in FPS on CPU compared to the original YOLOv3 algorithm. This improved algorithm ensures the detection accuracy based on a significant increase in detection speed, which reflects the theoretical and application value of the research. |
Year | DOI | Venue |
|---|---|---|
2022 | 10.1007/978-3-031-10989-8_35 | Knowledge Science, Engineering and Management |
Keywords | DocType | ISSN |
MobileNetv3, Object detection, YOLOv3, Lightweight target detection algorithm, CIoU | Conference | 0302-9743 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
5 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Fang Tong | 1 | 0 | 0.68 |
| Du Baoshuai | 2 | 0 | 0.68 |
| Xue Yunjia | 3 | 0 | 0.34 |
| Yang Guang | 4 | 0 | 0.68 |
| Jing-bo Zhao | 5 | 10 | 7.38 |