# FSIV P6-2 - Pollen classification - 2024 — Data

来源：https://www.kaggle.com/competitions/fsiv-p-6-2-pollen-classification-2024/data

采集时间：2026-10-22T09:50:47Z

A subset of the BAA500 dataset has been uploaded to Moodle. You will find three files: `train.tar.gz`, `valid.tar.gz` and `test.tar.gz`. It is recommended that you extract the three files in the `data` directory at the root of your project. The resulting file structure should resemble the following:

```
data
├── test
│   ├── 000040.png
│   ├── 000041.png
│   ├── 000070.png
              ...
├── test.csv
├── train
│   ├── 000000.png
│   ├── 000005.png
│   ├── 000011.png
              ...
├── train.csv
├── valid
│   ├── 000019.png
│   ├── 000039.png
│   ├── 000042.png
              ...
└── valid.csv
```

The dataset contains 24,000 training samples, 3,000 validation and 3,000 test samples. Each sample is a 128 x 128 grayscale image containing the (segmented) image of a single grain of pollen. While the train and validation samples are labelled, the samples in the test dataset are labelled as `unknown`.