Simple web demo of real-time Calorie Calculator using Yolov3 Object Detection model.
[Youtube Demo]

• Tensorflow 1
• CuDNN=1

Our team created an Object Detection dataset based on Classification dataset, "FoodDataSet"[1]. It is also available on Kaggle.

bound_process.py is used for specific annotation setting for yolov3.

All codes utilized in preprocessing is located in /preprocess.

We trained YOLOv3-tiny model for our custom dataset.

Our result demonstrated following result on our custom dataset.

You could reproduce our result by following command.

git clone https://github.com/ssumin6/ML-Calorie-Calculator.git
cd model_web
make

For individual image processing.

./darknet detector test data/food.data cfg/yolov3-tiny_obj.cfg final.weights {image_path}

[1] https://github.com/corona10/FoodDataSet
[2] https://github.com/pjreddie/darknet
[3] https://github.com/AlexeyAB/darknet
[4] https://github.com/neltia/flask-project