What is it?

Our mission was to utilize Image Processing techniques to understand Mars' layout and identify hazards for the rover's safety. With over 100,000 NASA-provided images, we developed a Machine Learning algorithm to analyze and detect hazards accurately. Most algorithms improved the model, but one decreased accuracy due to noise. Training the AI took around 4 hours with that algorithm, compared to 20-30 minutes without it.Our algorithm employed various techniques, including preprocessing, edge detection (Canny, Gaussian, Laplacian), morphological operations, and local binary patterns. Despite limitations in achieving 100% accuracy, we achieved significant progress by leveraging a large sample size, iterating the data, and using oversampling techniques.