An application to help learn flat shapes covering properties, concepts of perimeter and area. Presented using Artificial Intelligence technology through a guessing game.
The dataset spans six classes: circle, kite, parallelogram, square, trapezoid, triangle. Currently (version 2), the dataset consists of 24,000 images.
- Circle: Train(3,000); Val(500); Test(500)
- Kite: Train(3,000); Val(500); Test(500)
- Parallelogram: Train(3,000); Val(500); Test(500)
- Square: Train(3,000); Val(500); Test(500)
- Trapezoid: Train(3,000); Val(500); Test(500)
- Triangle: Train(3,000); Val(500); Test(500)
Collection methodology: Using automation with Python Image Library (PIL)
Dataset maker: https://github.com/reevald/MathEdu/blob/main/dataset_maker_v2.ipynb
More info: https://www.kaggle.com/reevald/geometric-shapes-mathematics
| TEST ACCURACY with 6/1/1 ~ train/val/test | |
|---|---|
| Model V1 with Dataset V1 | Model V2 (improve NN) with Dataset V2 |
| 99.94% | 99.97% |
