# EGCO486-Final_Exam-2025 — Data

来源：https://www.kaggle.com/competitions/egco-486-final-exam-2025/data

本地 Meta Kaggle 快照；不是本次实时页面。

# Dataset Description

![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2561061%2Fc6ea5cbae316c511d4d125b4f8de6b6a%2Ficon.png?generation=1763621423648230&alt=media)
- The image size is (128, 128) pixel.
- The possible regular polygons are triangles, squares, pentagons and hexagons. A classifier should be able to predict what's the shape in an unseen image.
- The possible colors for foreground and background are ["white", "gray", "black", "red", "green", "blue", "yellow", "orange", "purple", "pink"], but they are slightly altered for every image. These color labels can be predicted using a multiclass classifier model.
- The bounding circle is always within the image boundaries.
- The rotation can be any degree between 0 and 360.

<h2>File descriptions</h2>
<ul>
<li><b>sample-submission.csv</b> - a sample submission file in the correct format</li>
<li><b>starter_code.ipynb</b> - a python code to process a feature extraction, model and submission file</li>
<li><b>train</b> - the training set contain 4 folders (hexagon, pentagon, square, triangle) contains 1.5k images</li>
<li><b>test</b> - the test set image contains around 1,393 images</li>
</ul>