Human-Segmentation-with-Dynamic-LiDAR-Data

Sample code for Human Segmentation with Dynamic LiDAR Data.

Introduction

This is the official code of Human Segmentation with Dynamic LiDAR Data.

Requirement

Keras 2.2.4

Quick start

Install

git clone https://github.com/tzhong518/Human-Segmentation-with-Dynamic-LiDAR-Data

Data preparation

Please download data from Learning-Based Human Segmentation and Velocity Estimation Using Automatic Labeled LiDAR Sequence for Training.

You can download weight of 4-frame model from here

Train and test

python Train.py \
       --model model's name \
       --gpu 0 \
       --frame 04

python Test.py \
       --test Test \
       --network /path/to/model \
       --gpu 0 \
       --frame 04

Citation

If you find this work or code is helpful in your research, please cite:

Zhong T, Kim W, Tanaka M, et al. Human Segmentation with Dynamic LiDAR Data[J]. arXiv preprint arXiv:2010.08092, 2020.