Accuracy improvement in human detection using HOG features on train-mounted camera
Shintaro Saika, Saki Takahashi, Masara Takeuchi, Jiro Katto · 2016
Nowadays, researches on accident prevention using train-mounted cameras had been progressing. Our proposed method considers temporal continuity between frames by using motion vectors in addition to conventional thresholding on similarity values obtained by a human detection method using HOG features. Experiments show the effectiveness of our method as compared with a previous method using only HOG features.