Papers › Stacked Homography Transformations for Multi-View Pedestrian Detection

Stacked Homography Transformations for Multi-View Pedestrian Detection

1 Jan 2021ICCV 2021 10archive 2025-07-28

Liangchen Song, Jialian Wu, Ming Yang, Qian Zhang, Yuan Li, Junsong Yuan

Multi-view pedestrian detection aims to predict a bird's eye view (BEV) occupancy map from multiple camera views. This task is confronted with two challenges: how to establish the 3D correspondences from views to the BEV map and how to assemble occupancy information across views. In this paper, we propose a novel Stacked HOmography Transformations (SHOT) approach, which is motivated by approximating projections in 3D world coordinates via a stack of homographies. We first construct a stack of transformations for projecting views to the ground plane at different height levels. Then we design a soft selection module so that the network learns to predict the likelihood of the stack of transformations. Moreover, we provide an in-depth theoretical analysis on constructing SHOT and how well SHOT approximates projections in 3D world coordinates. SHOT is empirically verified to be capable of estimating accurate correspondences from individual views to the BEV map, leading to new state-of-the-art performance on standard evaluation benchmarks.

PaperPDF

Code

No code repository is listed for this paper in the archive or in Syntology's graph.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Multiview DetectionPedestrian Detection

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Multiview Detection CVCS SHOT F1_score (1m) 67.0 #3 of 6 Archive leaderboard report
Multiview Detection CVCS SHOT MODA (1m) 45.0 #3 of 6 Archive leaderboard report
Multiview Detection CVCS SHOT MODP (1m) 77.4 #3 of 6 Archive leaderboard report
Multiview Detection CVCS SHOT Precision (1m) 83.6 #3 of 6 Archive leaderboard report
Multiview Detection CVCS SHOT Recall (1m) 55.9 #3 of 6 Archive leaderboard report
Multiview Detection CityStreet SHOT F1_score (2m) 71.8 #4 of 5 Archive leaderboard report
Multiview Detection CityStreet SHOT MODA (2m) 53.5 #4 of 5 Archive leaderboard report
Multiview Detection CityStreet SHOT MODP (2m) 72.4 #4 of 5 Archive leaderboard report
Multiview Detection CityStreet SHOT Precision (2m) 91.0 #4 of 5 Archive leaderboard report
Multiview Detection CityStreet SHOT Recall (2m) 59.4 #4 of 5 Archive leaderboard report
Multiview Detection MultiviewX SHOT MODA 88.3 #8 of 9 Archive leaderboard report
Multiview Detection MultiviewX SHOT MODP 82.0 #8 of 9 Archive leaderboard report
Multiview Detection MultiviewX SHOT Recall 91.5 #8 of 9 Archive leaderboard report
Multiview Detection Wildtrack SHOT MODA 90.2 #8 of 10 Archive leaderboard report
Multiview Detection Wildtrack SHOT MODP 76.5 #8 of 10 Archive leaderboard report
Multiview Detection Wildtrack SHOT Recall 94.0 #8 of 10 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections