README.md

Multispectral Deep Neural Networks for Pedestrian Detection

Editted by Jingjing Liu, Rutgers University.

Code used in reproducing results in our paper Multispectral deep neural networks for pedestrian detection by Jingjing Liu, Shaoting Zhang, Shu Wang, and Dimitris N. Metaxas. BMVC 2016. [project link].

This repository is a folk of py-faster-rcnn offciel code, written by Ross Girshick. For how to install the required softwares and set up the code in right configuration, e.g., Caffe, pycaffe, please refer to their original README.md.

Download pretrained models

VGG16 model on caltech trained on Caltech pedestrian dataset.

VGG16 model on kaist (RGB input) trained on Kaist pedestrian dataset.

VGG16 model on kaist (multispectral input) trained on Kaist multispectral dataset.

Save these models to models/caltech/VGG16/, models/kaist/VGG16/, and models/kaist_fusion/VGG16/, respectively.

Run demos

Run sh ./run_demo.sh caltech for images from Caltech.

Run sh ./run_demo.sh kaist-color for images from Kaist.

Run sh ./run_demo.sh kaist-fusion for multispectral images from Kaist.

Caltech results

KAIST results

License

Our code is released under the MIT License (refer to the LICENSE file for details).

Citing our paper

If you find our work useful in your research, please consider citing:

@article{liu2016multispectral,
  title={Multispectral deep neural networks for pedestrian detection},
  author={Liu, Jingjing and Zhang, Shaoting and Wang, Shu and Metaxas, Dimitris N},
  journal={arXiv preprint arXiv:1611.02644},
  year={2016}
}