Abstract | ||
---|---|---|
Greedy-NMS inherently raises a dilemma, where a lower NMS threshold will potentially lead to a lower recall rate and a higher threshold introduces more false positives. This problem is more severe in pedestrian detection because the instance density varies more intensively. However, previous works on NMS don't consider or vaguely consider the factor of the existent of nearby pedestrians. Thus, we propose \heatmapname (\heatmapnameshort ), which pinpoints the objects nearby each proposal with a Gaussian distribution, together with \nmsname, which dynamically eases the suppression for the space that might contain other objects with a high likelihood. Compared to Greedy-NMS, our method, as the state-of-the-art, improves by $3.9%$ AP, $5.1%$ Recall, and $0.8%$ MR\textsuperscript-2 on CrowdHuman to $89.0%$ AP and $92.9%$ Recall, and $43.9%$ MR\textsuperscript-2 respectively.
|
Year | DOI | Venue |
---|---|---|
2020 | 10.1145/3394171.3413617 | MM '20: The 28th ACM International Conference on Multimedia
Seattle
WA
USA
October, 2020 |
DocType | ISBN | Citations |
Conference | 978-1-4503-7988-5 | 0 |
PageRank | References | Authors |
0.34 | 19 | 7 |
Name | Order | Citations | PageRank |
---|---|---|---|
Penghao Zhou | 1 | 0 | 1.01 |
Chong Zhou | 2 | 0 | 0.34 |
Pai Peng | 3 | 0 | 1.35 |
Junlong Du | 4 | 0 | 0.68 |
Sun Xing | 5 | 33 | 10.94 |
Xiaowei Guo | 6 | 0 | 0.34 |
Feiyue Huang | 7 | 226 | 41.86 |