IZ*Net is the definitive way to differentiate between the Jo Yuriz trio!
Ok, the astute WIZ*One will notice that I have mislabelled all three girls, so let's try again:
Much better!
All jokes aside, IZ*Net is just a fun project I worked on to gain some experience with deep learning and CNNs. You can read up about my adventure in the writeup I put together. (Even if you don't care about deep learning and CNNs, I would recommend giving a brief look at Appendix A. I use a segment tree to pick uniformly random objects!)
For a bit of a more technical background, IZ*Net consists of two networks: one to detect faces using the YOLO algorithm and another to recognize the faces.
If you'd like to look at code:
faceRecognition
- Is a module I used to develop the face recognition model.notebooks
- Contains my notebooks for this projectFaceDetection.ipynb
- Develops the face detection model using the YOLO algorithmFaceRecognition.ipynb
- Develops the face recognition modelIZ_NET.ipynb
- Develops the final systemObjectDetectionTest.ipynb
- Develops a rectangle detection model (used this to make sure I had implemented YOLO correctly)
misc
- Miscellaneous files (face/bounding box extraction scripts, scraper script, etc.)writeup
- All my LaTeX files, if for whatever reason you wanted to look at that