htyangs / Unsupervised-Histology-Clustering

Unsupervised Histology Clustering

Unsupervised Histology Image Clustering

How to use

(1) clone the github repository (include all the .py file in the directory)
(2) type sh download.sh to download model (best.pt)
(3) put the training image folder and testing image folder in the main directory
(4-a) execute train.py, it will output the model weight of each epoch
(4-b) execute inferece.py, it will output a file called result.csv

Train model

train.py:

first argument --data: input the folder of train and test

Example : python "/data1/home/8B07/Anthony/simsiam/train.py" --data /data1/home/8B07/Anthony/simsiam/

Export csv

Inference.py :

first argument --data: input the folder of train and test

second argument --model : input the model's directory name

Example : python "/data1/home/8B07/Anthony/simsiam/inference.py" --data /data1/home/8B07/Anthony/simsiam/ --model /data1/home/8B07/Anthony/simsiam/best.pt

Package requirements:

No specific package is required in the project

Code Similarity

Our model is based on Simsiam and SimTriplet, https://github.com/hrlblab/SimTriplet
Im my_model.py, Line 17-41 75-161 will be similar to the https://github.com/hrlblab/SimTriplet/blob/main/models/simsiam.py