Laser & Optoelectronics Progress, Volume. 61, Issue 8, 0837015(2024)
Few-Shot Object Detection Based on Association and Discrimination
Fig. 1. Basic structure of Faster R-CNN
Fig. 2. Structure of TFA two-stage fine-tuning method
Fig. 3. Decision boundary of TFA fine-tuning stage
Fig. 4. Decision boundary of association step
Fig. 5. Decision boundary of discrimination step
Fig. 6. Step of association stage
Fig. 7. Step of discrimination stage
Fig. 8. Dynamic R-CNN. (a) DLA; (b) DSL
Fig. 9. ECA module
Fig. 10. Prediction results of FSAD and TFA. (a) FSAD; (b) TFA
Fig. 11. Prediction results of differnet algorithms. (a) MPSR; (b) Retentive R-CNN; (c) DiGeo; (d) HTRPN; (e) FSAD
Fig. 12. Coexisting instances, left is semantic similarity, right is visual similarity
|
|
|
|
|
|
|
|
|
Get Citation
Copy Citation Text
Jianli Jia, Huiyan Han, Liqun Kuang, Fangzheng Han, Xinyi Zheng, Xiuquan Zhang. Few-Shot Object Detection Based on Association and Discrimination[J]. Laser & Optoelectronics Progress, 2024, 61(8): 0837015
Category: Digital Image Processing
Received: Jul. 5, 2023
Accepted: Aug. 22, 2023
Published Online: Apr. 16, 2024
The Author Email: Han Huiyan (hhy980344@163.com)