mantasu / glasses-detector

Glasses detection, classification and segmentation

https://mantasu.github.io/glasses-detector/

MIT License

55 stars 7 forks source link

TODO for v1.0.0 (Late February 2024 Update) #2

Closed mantasu closed 8 months ago

mantasu commented 1 year ago

Features and Improvements for v1.0.0

Data

[x] Update scripts to process dataset-by-dataset instead of task-by-task
[x] Find or add more datasets for classification, detection, and segmentation
[x] Write processing scripts, update preprocess.py CLI
[x] Upload the datasets as a single big one to Kaggle

Architecture

[x] Add prediction interface to handle predictions and conversion + saving to different formats
[x] Update base model to be extendable with glasses classifier, detector, and segmenter children
[x] Change class-based model instantiation to arg-based, create default models for each specification
[x] Require task for CLI, add size alias, detect device when None

Features

[x] Add detection dataset, tiny detection model, and binary detection wrapper
[x] Identify the best hyperparameters and set them as default for each wrapper
[x] Train models of different sizes and kinds for different tasks (36 in total)

Documentation

[x] Add a demo file and a collab notebook for quick examples
[x] Crop only the eye area in the tables, add more example images to showcase features
[x] Add the results for each model + their sizes, document training parameters
[x] Document data download and preprocessing steps
[x] Add type hints and documentation for the new classes/functions
[x] Add dataset + model architecture references