DenseBox: A Fully Convolutional Neural Net Framework for Object Detection

Introduction

DenseBox is an unified framework for object detection with a single network. You can use the code to train/evaluate a network for object detection task. For more details, please refer to our arXiv paper.

Directory Explanation

• caffe_densebox: The caffe code for running densebox detection. This folder is similar to caffe, with only core cpp detection code added.
• experiment: All the matlab/python/linux shell scripts for running experiments. If you want a start, please refer to experiment/kitti.
• paper: latex files for writing the densebox paper.