Short name:
SW_YoloX_det
Description:
Shift Window-YOLOX (SW-YOLOX) incorporates a novel Shift Window-Path Aggregation Feature Pyramid Network (SW-PAFPN) to integrate with the YOLOX detector, further enhancing feature extraction and the robustness of pedestrian detection.
Reference:
C.-Y. Tsai, R.-Y. Wang and Y.-C. Chiu, “SW-YOLOX: A Real-Time Pedestrian Detection System Based on Sliding Window-Mixed Attention Mechanism in YOLOX Lightweight Network,” Neurocomputing, accepted for publication, DOI: https://doi.org/10.1016/j.neucom.2024.128357
Last submitted:
June 23, 2023 (1 year ago)
Published:
July 03, 2023 at 11:24:27 CET
Submissions:
2
Project page / code:
Open source:
Yes
Hardware:
2.4 GHZ,24 Core
Runtime:
12.6 Hz
Benchmark performance:
Sequence | AP | MODA | MODP | FAF | TP | FP | FN | Rcll | Prcn | F1 |
MOT17Det | 0.90 | 86.4 | 83.0 | 1.0 | 105,082 | 6,082 | 9,482 | 91.7 | 94.5 | 93.1 |
Detailed performance:
Sequence | AP | MODA | MODP | FAF | TP | FP | FN | Rcll | Prcn | F1 |
MOT17-01 | 0.73 | 69.3 | 83.2 | 0.8 | 3,226 | 341 | 940 | 77.4 | 90.4 | 83.4 |
MOT17-03 | 0.90 | 94.6 | 83.3 | 2.2 | 67,686 | 3,274 | 436 | 99.4 | 95.4 | 97.3 |
MOT17-06 | 0.72 | 72.0 | 81.1 | 0.6 | 6,973 | 671 | 1,776 | 79.7 | 91.2 | 85.1 |
MOT17-07 | 0.90 | 90.0 | 82.8 | 0.8 | 8,709 | 376 | 551 | 94.0 | 95.9 | 94.9 |
MOT17-08 | 0.91 | 82.2 | 86.0 | 1.4 | 5,607 | 886 | 138 | 97.6 | 86.4 | 91.6 |
MOT17-12 | 0.82 | 78.8 | 85.2 | 0.2 | 4,471 | 223 | 923 | 82.9 | 95.2 | 88.6 |
MOT17-14 | 0.63 | 61.7 | 79.7 | 0.4 | 8,410 | 311 | 4,718 | 64.1 | 96.4 | 77.0 |
Raw data: