Skip to content

Latest commit

 

History

History
24 lines (15 loc) · 651 Bytes

README.md

File metadata and controls

24 lines (15 loc) · 651 Bytes

LecSD

Official github repository of Semantic Labels-Aware Transformer Model for Searching over a Large Collection of Lecture-Slides WACV 2023.

In order to setup the dataset do the following steps:

Download all the slides by running

cd datasets

pytthon 1_download_slides.py

It will save all the slide images with a unique ID.

The slide summary of each slide are given in the slide_summary.json file.

The semantiic informations are given in slide_summary.json file.

The train, validation, test split of slide images are in slide_summary.json file.

All the manually drawn sketch queries are included in the sketchQueries folder