YOLO-World Open Vocabulary test using Ultralytics ❤️🎉

:muscle: Steps involved

:diamond_shape_with_a_dot_inside: No model fine-tuning is needed, I have used the pretrained YOLOv8s-world model.
:diamond_shape_with_a_dot_inside: You can set class names manually after loading the model.
:diamond_shape_with_a_dot_inside: I used the video from the OpenAI Sora demo for testing.

:star2: classes = [“Elephant walking”, “Elephant standing”]

Learn more :arrow_right: YOLO-World (Real-Time Open-Vocabulary Object Detection) - Ultralytics YOLO Docs

Complete code is available in the comments :point_down:

Elephant standing Elephant walking

Code Snippet :robot_dance:

# pip install ultralytics

from ultralytics import YOLOWorld
model = YOLOWorld("yolov8s-world.pt")

model.set_classes( ["Elephant walking", "Elephant standing"]) # Define custom classes names

results = model.predict("Path/to/video/file.mp4")