To run on your machine, first create an environment by running
conda create --name NAME --file=requirements.txt anaconda
Everything is configured so that you can simply run
python3 engine.py
to see it in action.
Basic stats about data generated by the engine to discuss in the results section.
Folder for data. Note that the subfolders data/boxes
and data/images
are reserved for intermediate output. It's recommended that videos are placed in data/videos
.
You're here right now.
Deprecated python notebooks used for prototyping the final models and scripts. The file structure has changed a few times, so some edits may be needed to make notebooks functional.
Recommended folder to store output.
classify.py
- Script that, after pre-processing, performs feature extraction and solicits input from the user to label the result data.engine.py
- The Engine running our system. Handles file input and progress monitoring.preprocess.py
- Performs initial OCR to detect and merge high-probability bounding boxes for text.process.py
- Once groups are classified, performs the work of extracting the actual sentences and formatting in a friendly way in the output.vid2frames.py
Simpleffmpeg
wrapper that converts a video to frames at a given rate.