A 48-hour data challenge within a team of 3 students.
Were were provided with a dataset of 2350 labelled images (10% were noisy images) and we had to train a model to detect cars of 100 different models in order to calculate the carbon footprint of the whole car fleet of a company: frame each vehicle with a box and identify the car model, then deduce the carbon footprint of the car in the image. We were given a test dataset of 1500 images 48 hours after to evaluate our model.