kyegomez/odin

SOTA Classification at scale for UAVs, Drones, and much more

Odin

Super Fast and super cheap object detection at massive scale in less than 10 lines of code!

Appreciation

• Lucidrains
• Agorians

Install

pip install vodin

Usage

Here are three examples demonstrating the usage of the Odin class from your provided code:

Example 1: Basic Usage

# Import the necessary modules and classes
from odin import Odin

# Initialize the Odin object with paths and thresholds
odin = Odin(
    source_weights_path="yolo.weights",
    source_video_path="input_video.mp4",
    target_video_path="output_video.mp4",
    confidence_threshold=0.3,
    iou_threshold=0.7
)

# Run the object to process the video
odin.run()

Example 2: Custom Parameters

# Import the necessary modules and classes
from odin import Odin

# Initialize the Odin object with custom parameters
odin = Odin(
    source_weights_path="custom_yolo.weights",
    source_video_path="input_video.mp4",
    target_video_path="output_video.mp4",
    confidence_threshold=0.5,
    iou_threshold=0.6
)

# Run the object to process the video
odin.run()

Example 3: Advanced Usage

# Import the necessary modules and classes
from odin import Odin

# Initialize the Odin object with paths and thresholds
odin = Odin(
    source_weights_path="yolo.weights",
    source_video_path="input_video.mp4",
    target_video_path="output_video.mp4",
    confidence_threshold=0.3,
    iou_threshold=0.7
)

# Customize further configurations if needed
odin.tracker.set_max_distance(50)
odin.box_annotator.set_box_color((0, 255, 0))
odin.model.set_device("cuda")

# Run the object to process the video
odin.run()

Architecture

• Odin utilizes YoloV7, weights can be downloaded here

License

MIT