Shasin-Finder

Search for images in a folder or photo library by tags, objects present or photo description. A computer vision approach to help finding images based on how people actually remember their photos.

Requirements and dependencies

Dependencies installation under pip package manager

python3 -m pip install -r requirements.txt
# If requirements.txt fails: 
# Pytorch: https://pytorch.org/get-started/locally/
# Detectron2: https://github.com/facebookresearch/detectron2/blob/master/INSTALL.md

Inference

python3 main.py --folderPath './input' --labelFile labels.csv

CLI Arguments:

• '--folderPath' : Folder to search for images Default = './input'
• '--modelDetection' : Filename of weights associated with detection
• '--cfgPath' : Path to detectron model cfg file relative to 'detectron2/model_zoo'
• '--labelFile' : Path to model labels (csv file)

Examples:

• Photo was taken in winter:

• Nature, there were many animals

• That cat was wearing a tie

Demo Video:

https://www.youtube.com/watch?v=oW-_hs5r68s&feature=youtu.be&ab_channel=UtsavMaskey