This project was implemented by Ian Stähli and Miro Voellmy in the frame of our semester thesis at Prof. Otmar Hilliges Advanced Interactive Technologies Lab at ETH Zurich from 03/17 - 07/17.
The goal of the self initiated project was to implement a pipeline which can estimate the state of the stove top and its objects such as the operator, the pans, the tools, the plates and the food.
A custom dataset was created in order to train a object detection, recognition pipeline as well as a gesture recognition pipeline. The dataset contained 3 dishes, which were prepared multiple times on two different stoves. The project was implement in Python 3 using OpenCV and scikit-learn.
Feel free to check out the demonstration of the pipeline in the videos and the source code on GitHub.