Laser & Optoelectronics Progress, Volume. 56, Issue 20, 201501(2019)
Posture-Guided and Multi-Granularity Feature Fusion for Person Reidentification
Fig. 1. Network architecture
Fig. 2. PAF network architecture
Fig. 3. PAF rendering
Fig. 4. MGTN architecture
Fig. 5. Diagram of part segmentation. (a) 4 parts; (b) 6 parts
Fig. 6. Effect of descriptor H with different resolutions at M=4,6 (Market1501). (a) Rank-1; (b) mAP
Fig. 7. Comparison of H and G at M=6 (Market1501). (a) Rank-1; (b) mAP
Fig. 8. Effect of descriptor G with different resolutions at M= 4,6 (CUHK03-NP). (a) Rank-1; (b) mAP
Fig. 9. Effect of descriptor G with different resolutions at M=6 (DukeMTMC-reID). (a) Rank-1; (b) mAP
|
|
|
|
|
|
|
|
|
|
Get Citation
Copy Citation Text
Liang Zhang, Jin Che. Posture-Guided and Multi-Granularity Feature Fusion for Person Reidentification[J]. Laser & Optoelectronics Progress, 2019, 56(20): 201501
Category: Machine Vision
Received: Mar. 20, 2019
Accepted: Apr. 26, 2019
Published Online: Oct. 22, 2019
The Author Email: Che Jin (koalache@126.com)