Extracting the Clicks from CSV

Annotation sessions saved by pyclickimage can be loaded directly from a CSV file without opening the graphical interface.

For most use cases, the recommended approach is to use the pyclickimage.read_session_csv() function. It provides a simple way to access the annotations as a Python dictionary or as a pandas DataFrame.

Reading a session

To extract the clicks from a saved session:

import pyclickimage

session = pyclickimage.read_session(
    "session.csv"
)

The default output is a nested dictionary organized as:

{
    "image_1.png": {
        "default": [
            (276.5, 97.0),
            (242.2, 109.8),
        ],
        "target": [
            (317.0, 144.5),
        ],
    },

    "image_2.png": {
        "default": [
            (50.0, 25.5),
        ],
    },
}

Accessing clicks

Clicks can then be accessed by image and group:

clicks = session["image_1.png"]["default"]

print(clicks)

# Output:
# [(276.5, 97.0), (242.2, 109.8)]

Using the core classes

For advanced applications, the lower-level classes can be used directly.

AnnotationSession allows loading and manipulating a complete annotation session, while ClickManager manages the annotations associated with a single image.

These classes are mainly intended for developers extending the package. read_session_csv is recommended for simple extraction and analysis tasks.