lihengMo's repositories
group-people-in-images
This is a project I did during the internship in IBM Research Tokyo. I used a CNN framework to estimate the human pairwise relation based on the humans location and head/face direction features. With using the estimation of the pairwise matrix calculated by the Pair-Net, using a customized spectral clustering to cluster the humans into different groups.
Language:Python000
SCUT_FIR_Pedestrian_Dataset
A new benchmark dataset and baseline for on-road FIR pedestrian detection