This is a short implementation of the YOLOv3 Framework to train a model to detect and localize the faces of me and 4 of my friends (DJ, Gab, Nicolo, Nuy) using Pytorch backend and Roboflow for image labelling and annotation. The YOLOv3 framework is a Deep Learning framework originally developed for fast object localization tasks (with hundreds to thousands of active classes). For more information about YOLOV3, check out this great blog.
LOGS
- Model trained using 25 images from each friend class under 50 epochs.
- Average model performance as expected
- Local machine (laptop) does not have sufficient computing power and memory to perform larger scale training
ASSESSMENT
- Algorithm and preprocessing steps works and can be improved with more resources
SAMPLE OUTPUT
