Welcome, you are visitor number 
Overview
This project was done at IvLabs to implement face-recognition algorithms from scratch, and further plan to deploy as face recognition based door lock. During this, we implemented Triplet Network and FaceNet algorithms with ResNet as the backbone architecture implemented from scratch and performed one-shot and few-shot learning on different datasets.
Triplet Network
|
Loss function:
|
Results
1] MNIST
- Implemented as a warmup for the project to gain an understanding of one-shot learning and siamese nets.
- The model was trained only on 100 images of classes 0,1,2. Images of rest classes i.e 3-9 were kept hidden during the training phase.
- The evaluation was done on a 10-way 1-shot basis, wherein the support set had only one sample from each class.
|
|
|
2] ORL (Datasets of Faces)
- Trained using ResNet backbone implemented from scratch. During the training phase, experimented with different triplet mining methods i.e offline triplet mining and online triplet mining.
|
|
3]LFW(Labelled Faces in the Wild)
- Implemented the paper FaceNet: A Unified Embedding for Face Recognition and Clustering on LFW Dataset. Notes for the same paper can be found here.
Created a Dataset class for LFW and contributed it to Torchvision.
Wandb was used throughout this part of the project for metric tracking, hyperparameter tuning, sweeps, visualization, etc.
4] Lab Members
For more results and information checkout:
Team
- Muhammed Abdullah
- Khurshed Fitter
- Rushika Bhagwatkar
