Pedestrian Detection Combining RGB and Dense LIDAR Data

Pedestrian Detection Combining RGB and Dense LIDAR Data
C. Premebida¹, J. Carreira^1,2, J. Batista¹ and U. Nunes¹
¹ISR, DEEC, Univ. of Coimbra. ²UC Berkeley

Dense depth-map upsampling method using only LIDAR
3D Velodyne and camera fusion (experiments: KITTI dataset)
Def. Part-Models + SVM-based rescoring strategy
High performance on KITTI benchmarking

Results on the KITTI dataset (Fusion-DPM): http://www.cvlibs.net/datasets/kitti/eval_object.php
Paper: pdf (825KB) IEEE, IROS' 14
\cite: bib

Implementations
Notes:
our solution depends on Discriminatively trained deformable part models

R. B. Girshick, P. F. Felzenszwalb, and D. McAllester, “Discriminatively Trained Deformable Part Models, Release 5”
LIBSVM is also necessary libsvm-3.12

README, .m and .cpp (mex) files: download