Here, the challenge is to classify each driver's behavior. Are they driving attentively, wearing their seatbelt, or taking a selfie with their friends in the backseat?

• scikit-learn
• Python >= 2.7
• Caffe
• matplotlib
• OpenCV >= 3.0

1. Download the data : https://www.kaggle.com/c/state-farm-distracted-driver-detection
2. Data preparation:

If "stateFarm_train.txt" and "stateFarm_test.txt" are not generated, run:

$ python prepare_data.py

3. Train RGB model

$ ./run_singleFrame_RGB.sh

Make sure to change the "root_folder" param in "CNN.prototxt" as needed.

4. Evaluate on test

$ python classify_test.py

This script classifies the test imgs and fill the submission csv for Kaggle.

5. Evaluation

Plot train/test accuracy curve:

$ python generate_metrics.py

Confusion matrix plot:

$ python generate_confusion_matrix.py

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change. Please make sure to update tests as appropriate.

Obs.: The models are highly based on ([LRCN])(https://github.com/LisaAnne/lisa-caffe-public/tree/lstm_video_deploy/examples/LRCN_activity_recognition) repository

Fell free to contact Marcos Teixeira if you have any questions.