Self-Driving Car Engineer Diary — 7

Tue, 14/Mar/2017

Vehicle Detection and Tracking

The Computer Vision learning continued in project 5. Training a machine learning Support Vector Machine (SVM) classifier with selected Histogram of Oriented Gradients (HOG), spatial (32 x 32) and colour features (3 channels ; YCrCb) to arrive at a Test Accuracy of 0.995.

Sliding Window technique was shown and the use of heatmaps to help limit the number of false-positive detections of cars.

Results were ‘fun’ …

… but everyone realised how brittle this was and wanted to replace SVM with their favourite Deep Learning model. I’m keen to apply the Single-shot Multi-Box Detector model in a future application.

Term 1 of 3 Retrospective

Amazing projects. Steep learning curve. Strong student community. Incredibly supportive and adaptive Udacity staff. Be prepared to commit 2–3 times estimated 10 hours per week to complete Term 1 successfully as projects encourage experimentation. Now to catch-up on sleep before the start of Term 2 on 24/Mar/2017.