PaddleX Image Classification Task Module Data Annotation Tutorial¶

This document will introduce how to use the Labelme annotation tool to complete data annotation for image classification related single models. Click on the above link to refer to the homepage documentation for installing the data annotation tool and viewing detailed usage procedures.

1. Labelme Annotation¶

1.1 Introduction to Labelme Annotation Tool¶

Labelme is a Python-based image annotation software with a graphical interface. It can be used for tasks such as image classification, object detection, and image segmentation. In instance segmentation annotation tasks, labels are stored as JSON files.

1.2 Labelme Installation¶

To avoid environment conflicts, it is recommended to install in a conda environment.

conda create -n labelme python=3.10
conda activate labelme
pip install pyqt5
pip install labelme

1.3 Labelme Annotation Process¶

1.3.1 Prepare Data for Annotation¶

Create a root directory for the dataset, such as pets.
Create an images directory (must be named images) within pets and store the images to be annotated in the images directory, as shown below:

Create a category label file flags.txt for the dataset to be annotated in the pets folder, and write the categories of the dataset to be annotated into flags.txt line by line. Taking the flags.txt for a cat and dog classification dataset as an example, as shown below:

1.3.2 Start Labelme¶

Navigate to the root directory of the dataset to be annotated in the terminal and start the labelme annotation tool.

cd path/to/pets
labelme images --nodata --autosave --output annotations --flags flags.txt

* flags creates classification labels for images, passing in the path to the labels. * nodata stops storing image data in JSON files. * autosave enables automatic saving. * output specifies the storage path for label files.

1.3.3 Start Image Annotation¶

After starting labelme, it will look like this:

* Select the category in the Flags interface.

After annotation, click Save. (If output is not specified when starting labelme, it will prompt to select a save path upon the first save. If autosave is specified, there is no need to click the Save button).

* Then click Next Image to annotate the next image.

After annotating all images, use the convert_to_imagenet.py script to convert the annotated dataset to the ImageNet-1k dataset format, generating train.txt, val.txt, and label.txt.

python convert_to_imagenet.py --dataset_path /path/to/dataset

dataset_path is the path to the annotated labelme format classification dataset.

The final directory structure after organization is as follows:

2. Data Format¶

The dataset defined by PaddleX for image classification tasks is named ClsDataset, with the following organizational structure and annotation format:

dataset_dir    # Root directory of the dataset, the directory name can be changed
├── images     # Directory for saving images, the directory name can be changed, but note the correspondence with the content of train.txt and val.txt
├── label.txt  # Correspondence between annotation IDs and category names, the file name cannot```bash
classname1
classname2
classname3
...

Modified label.txt:

```bash 0 classname1 1 classname2 2 classname3 ...