TCRNet-1 Code and Files

This repository contains the code and files for the TCRNet-1 model. For detailed instructions, refer to the PDF in the repository.

TCRNet-1 detects small vehicular targets in infrared imagery by maximizing the target-to-clutter ratio, improving detection accuracy in challenging cluttered environments with limited target pixels.

Table of Contents

• Overview
• Setup
• Generating Chips
• Training TCRNet-1
• Running and Evaluating TCRNet-1

Overview

TCRNet-1 is designed to enhance the detection of small vehicular targets in infrared imagery. It aims to maximize the target-to-clutter ratio, which is essential for accurate detection in environments where the targets are surrounded by clutter.

Setup

To set up the environment for TCRNet-1, follow these steps:

1. Install Anaconda:

   wget https://repo.anaconda.com/archive/Anaconda3-2020.07-Linux-x86_64.sh
   bash Anaconda3-2020.07-Linux-x86_64.sh

2. Create and Activate Conda Environment:

   conda env create -f environment.yml
   conda activate tcr

Generating Chips

Ensure that your data directory has GT and IMAGES subdirectories. For the ATR Database, atrdb_gt contains the appropriate GT directory.

Directory Structure

/path/to/processed/data
├── GT
│   ├── cegr01923_0001.txt
│   ├── cegr01923_0002.txt
│   └── ...
└── IMAGES
    ├── cegr01923_0001_0001.png
    ├── cegr01923_0001_0002.png
    ├── ...
    ├── cegr01923_0002_0001.png
    └── cegr01923_0002_0002.png

Running Chip Generation

In Python:

import ds_build
ds_build.build_ds('seqlists/trainlistv3.txt', 'data/exp1', datapath='/path/to/processed/data', skip=30)

Or via command line:

python ds_build.py --seqlist seqlists/trainlistv3.txt --outpath data/exp1 --datapath /path/to/processed/data --skip 30

Training TCRNet-1

Run the following command to train TCRNet-1:

python train_tcr.py --chippath ./data/exp1 --weightpath ./weights_filters/exp1

Running and Evaluating TCRNet-1

To run and evaluate the model, use the following command:

python validate_tcr.py --seqlist data/trainlist.txt --weightpath ./weights_filters/exp1 --datapath /path/to/processed/data --skip 30