# A02025-Medical-Image-Segmentation — Data

来源：https://www.kaggle.com/competitions/a0-2025-medical-image-segmentation/data

采集时间：2026-10-19T07:45:32Z

The dataset consists of medical images organized into two main folders: **Train** and **Test**. The structure and usage of each folder are as follows:  

- **Train** folder: Contains 1,087 images and their corresponding segmentation masks, organized into two subfolders:  
  - **Image**: Contains 1,087 input medical images in `.jpg` format.  
  - **Mask**: Contains 1,087 segmentation masks in `.png` format, where each mask corresponds to an image in the **Image** folder.  
  - Participants can split this dataset into training and validation sets as needed for model development and fine-tuning.  

- **Test** folder: Contains 192 medical images in `.jpg` format without corresponding masks. These images are used for both the **public test** and the **private test**:  
  - **Public Test**: Results from the public test will be displayed on the public leaderboard after submission.  
  - **Private Test**: Participants must select one of their previous submissions for evaluation on the private test. The final ranking will be determined based on the private test, and results will be displayed on the private leaderboard after the contest concludes.  

This dataset provides labeled training data for model development and unlabeled test data for final evaluation, challenging participants to accurately segment unseen images.