Dataset¶
This page describes the expected format for datasets used when training models with fast-plate-ocr.
Expected File Format¶
Your dataset should be provided as a CSV file with the following structure:
| Column Name | Type | Description |
|---|---|---|
image_path |
str |
Relative path to the license plate image |
plate_text |
str |
Ground-truth text on the plate (no padding) |
Relative Paths
Image paths in the CSV are resolved relative to the location of the CSV file, not the working directory.
Dataset Structure Example¶
The dataset should include at least a CSV file (for training and validation) and a folder with corresponding plate images.
dataset/
├── train/
│ ├── annotations.csv
│ └── images/
│ ├── 00001.jpg
│ ├── 00002.jpg
│ ├── 00003.jpg
│ ├── 00004.jpg
│ └── 00005.jpg
└── val/
├── annotations.csv
└── images/
├── 00006.jpg
└── 00007.jpg