Causal Localization Network for Radar Human Localization with micro-Doppler signature

Environment

• python 3.7
• torch 1.13.1
• CUDA 12.2

Setting

conda env create --file environment.yaml

Download IDRad data

You can download IDRad dataset below two links

• IDRad_original

• IDRad

Option: download IDRad-TBA (Temporal Boundary Annotation) dataset below link

• IDRad-TBA

Reproduce IDRad-TBA dataset for Radar Human Localization

python make_data.py

Prepare 'IDRad' folder for IDRad dataset

Prepare causal mask and spasity mask for Causal Localization Network (CLNet)

python make_mask.py

The outputs are 'causal_mask.npy' and 'sparsity_mask.npy'

Train CLNet for radar human localization task

python train.py

Inference only is also possible by commenting the training in the code.

Reference

Dual-scale Doppler Attention for Human Identification. Link

Indoor Person Identification with Radar Data. Link