Flow-Guided Feature Aggregation for Video Object Detection

ICCV 2017 Xizhou ZhuYujie WangJifeng DaiLu YuanYichen Wei

Extending state-of-the-art object detectors from image to video is challenging. The accuracy of detection suffers from degenerated object appearances in videos, e.g., motion blur, video defocus, rare poses, etc... (read more)

PDF Abstract

Evaluation Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK COMPARE
Video Object Detection ImageNet VID FGFA + Seq-NMS MAP 80.1% # 3
Video Object Detection ImageNet VID FGFA + Seq-NMS runtime (ms) 954 # 1