Example Codes and Utilities

Checkout this Github Repository for the most updated versions: https://github.com/XailientPublic/example_scripts

Scripts for inferencing

Split a video into frames for training data

File: extract_frames_from_video.py

This tool allows developers to quickly split a video into frames, with the aim in create training data for model production.

This python script allows developers to either:

Go through each frame and manually save the frames you want (default).

$ python3 extract_frames_from_video <path_to_video>

Save every ith frame in the video.

$ python3 extract_frames_from_video --every <integer> <path_to_video>

Conversion of different annotation formats

File: convert_label_to_xailient_format.py

A script to convert different annotations formats to the format required by the Xailient Console.

Currently supported formats for this conversion script are pascalvoc/labelimg, labelme, coco, and yolo formats.

Choices for input_format argument are 'voc', 'coco', 'labelme', 'yolo'

For annotations present in a single file (e.g. COCO), input_path represents the path to the JSON file.
While for separate annotations for each image (e.g. Pascal VOC, yolo, labelme), input_path represents the path to the folder where the annotations reside.

The output_path is the path and name of the converted xailient annotations.

Examples of usage:

$ python3 convert_label_to_xailient_format.py --input_path example/coco_annotations.json --input_format coco --output_path example/xailient_labels.csv