simsiam-pytorch

This repo contains:

• Minimal PyTorch Implementation of self-supervised SimSiam training from "Exploring Simple Siamese Representation Learning" by Chen et al.
• Implementation of Metric Learning

Install dependencies

pip install -r requirements.txt

To train metric model

B1. Create dataset that has the following format:
 train/
   1/
      .png
   2/
      .png
   ....
 eval/
   1/
      .png
   2/
      .png
      ....
 query/
   1/
      .png
   2/
      .png

B2. Modify parameters in the config file train_metric_learning.json such as: data.path, model.backbone, v.v B3. Run the following command python3 train_metric.py --cfg CONFIG_PATH

To get embedding vectors of img dataset

B1. Modify parameters in the config file get_metric_embedding.json B2. Run the following command python3 get_metric_mebedding.py --cfg CONFIG_PATH

Link to pretrained Xception on synthesis dataset link

Todo:

[x]: Implement metric learning model
[]: Optimize