TTHF

The code of Text-Driven Traffic Anomaly Detection with Temporal High-Frequency Modeling in Driving Videos Rongqin Liang, Yuanman Li, Jiantao Zhou, and Xia Li

Our TTHF framwork:

Installation

Dependencies

Python 3.8
pytorch 2.0.1
cuda 11.3
Ubuntu 20.04
RTX 3090
Please refer to the "requirements.txt" file for more details.

Preparing

First, prepare the data for training or testing by:

cd TTHF/datasets
python ./extract_samples.sh

Note that you need to modify the corresponding path.

Training

Users can train the TTHF models on DoTA dataset easily by runing the following command:

cd TTHF
python3 ./main.py \
        --train \
        --lr_clip 5e-6 \
        --wd 1e-4 \
        --epochs 15 \
        --batch_size 128 \
        --dataset DoTA \
        --gpu_num 0 \
        --height 224 \
        --width 224 \
        --normal_class 1 \
        --eval_every 1000 \
        --base_model 'RN50' \
        --general \
        --fg \
        --hf \
        --aafm \
        --other_method 'TDAFF_BASE' \
        --exp_name 'TDAFF_BASE_RN50'

Note that you need to modify the corresponding path.

Inference

Users can test the TTHF models (Extraction code: rb9k) on DoTA or DADA-2000 dataset easily by runing the following command:

For DoTA:

python3 ./main.py \
        --evaluate \
        --batch_size 128 \
        --dataset DoTA \
        --gpu_num 0 \
        --height 224 \
        --width 224 \
        --normal_class 1 \
        --eval_every 1000 \
        --base_model 'RN50' \
        --general \
        --fg \
        --hf \
        --aafm \
        --other_method 'TDAFF_BASE' \
        --exp_name 'TDAFF_BASE_RN50'

For DADA-2000:

python3 ./main.py \
        --evaluate \
        --batch_size 128 \
        --dataset DADA \
        --gpu_num 0 \
        --height 224 \
        --width 224 \
        --normal_class 1 \
        --eval_every 1000 \
        --base_model 'RN50' \
        --general \
        --fg \
        --hf \
        --aafm \
        --other_method 'TDAFF_BASE' \
        --exp_name 'TDAFF_BASE_RN50'

User can also see "tdaff_base_script.sh" for more training and testing commands.

Note that our project is developed based on the code of Learning Transferable Visual Models From Natural Language Supervision. The relevant pre-trained models can be downloaded from the official website.

Citation

If you found the repo is useful, please feel free to cite our papers:

@article{10504300,
    title={Text-Driven Traffic Anomaly Detection with Temporal High-Frequency Modeling in Driving Videos},
    author={Rongqin Liang and Yuanman Li and Jiantao Zhou and Xia Li},
    journal={IEEE Transactions on Circuits and Systems for Video Technology},
    year={2024},
    doi={10.1109/TCSVT.2024.3390173}
}

Blessinglrq / TTHF