The project was made for my practice at University where I'm currently studying. It was a machine learning task, where I needed to classify supermarket product images into 40 categories. For this purpose I used the following:

• Data analysis tools: pandas, numpy, matplotlib
• Machine learning libraries: tensorflow, scikit-learn
• Jupyter notebooks for my work presentation
• Web scraping to collect the images from more than 50 supermarkets worldwide - requests, beautifulsoup, selenium

For solving the image classification problem I used deep convoluted networks as well as transfer learning. After training eight models I chose the best of them and combined them using ensemble methods.
The resulting accuracy was 85%. The example of images is shown below: