Gated Channel Attention Mechanism YOLOv3 Network for Small Target Detection
In order to solve the problem of low recognition rate and high missed rate in current target detection task, this paper proposes an improved YOLOv3 algorithm based on a gated channel attention mechanism (GCAM) and adaptive up-sampling module. Firstly, darknet-53 is used as the backbone network to ex...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2022-01-01
|
| Series: | Advances in Multimedia |
| Online Access: | http://dx.doi.org/10.1155/2022/8703380 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In order to solve the problem of low recognition rate and high missed rate in current target detection task, this paper proposes an improved YOLOv3 algorithm based on a gated channel attention mechanism (GCAM) and adaptive up-sampling module. Firstly, darknet-53 is used as the backbone network to extract image basic features. Secondly, an adaptive up-sampling module is introduced to expand the low-resolution convolutional feature images, which effectively enhances the fusion efficiency of the convolutional feature images at different scales. Finally, GCAM is added to improve the network’s feature expression and detection capability for small targets before the three-scale channels output the prediction results. The results show that the improved method can adapt to multiscale target detection tasks in complex scenes and reduce the missing rate of a small target. |
|---|---|
| ISSN: | 1687-5699 |