This is an end-to-end pipeline that uses scikit-learn to train a linear SVM (support vector machine) on HOG (histogram of oriented gradients) features extracted from images and perform inference on a video. The SVM was trained on vehicles with the aim of detecting vehicles in a dashcam video.