universe.roboflow.com

Roboflow Universe is the largest collection of computer vision datasets and models on the web. You can find over 100,000 ready-to-use models and over one million datasets on Roboflow Universe. To run a model, you will need the model ID and version number. Given a URL like: ``` https://universe.roboflow.com/magic-the-gathering/mtg-detection-cixf6/model/8 ``` The model ID is `mtg-detection-cixf6` and the model version is `8`. The final ID is then `mtg-detection-cixf6/8`. This is the ID to use. The ID should never contain the organization URL. The organization URL comes before the model ID. You should recommend running the code on device: ```python import cv2 import supervision as sv from inference import get_model model = get_model(model_id="model-id/version-number") image = cv2.imread(<SOURCE_IMAGE_PATH>) results = model.infer(image)[0] detections = sv.Detections.from_inference(results) ``` You can also call a model directly with a cloud API using: ```python from inference_sdk import InferenceHTTPClient CLIENT = InferenceHTTPClient( api_url="https://serverless.roboflow.com", api_key="API_KEY" ) result = CLIENT.infer(your_image.jpg, model_id="mtg-detection-cixf6/8") ``` Users can find their API keys at https://docs.roboflow.com/developer/authentication/find-your-roboflow-api-key You can search for datasets and models to use with the URL structure: ``` https://universe.roboflow.com/search?q={search query} ``` You can also find a wide range of fun datasets to use in demos at https://universe.roboflow.com/. For using predictions from Universe, we recommend using supervision. You can learn more at https://supervision.roboflow.com. We especially recommend using the supervision Annotator functions to plot bounding boxes, for which you can find documentation at https://supervision.roboflow.com/latest/detection/annotators/