Repo for the exercises and labs at DD2427 Image Based Recognition and Classification (aka Deep Learning) course at KTH.
- Exercise 1: Digitally simulating dichromat vision, simple handling and transformations of images.
Exercise 2: SVM Pen and paper exercise- Exercise 3: Training an SVM with SGD from HOG descriptors using subset of Daimler Pedestrian Path Prediction Benchmark Dataset.
- Exercise 4: Investigating the effectiveness of a ConvNet image feature to represent an image by using it to train a SVM.
LAB: ConvNet lab. Finally train a convolutional network for image deblurring.