Measure objects' size and shape. These models are ready to go; often with pre-trained weights and exports available for mobile or server-side inference.
If you're more interested in deploying a model without code, check out our Roboflow Deploy product.
Model Size:
MB
Parameters:
Top FPS:
Architecture:
Model Size:
MB
Parameters:
Top FPS:
Architecture:
Model Size:
MB
Parameters:
Top FPS:
Architecture:
Model Size:
MB
Parameters:
Top FPS:
Architecture:
Model Size:
MB
Parameters:
Top FPS:
Architecture:
Model Size:
MB
Parameters:
Top FPS:
Architecture:
Model Size:
MB
Parameters:
Top FPS:
Architecture:
Model Size:
MB
Parameters:
Top FPS:
Architecture:
Model Size:
MB
Parameters:
219 million
Top FPS:
Architecture:
Transformers
Model Size:
MB
Parameters:
Top FPS:
Architecture:
Model Size:
MB
Parameters:
Top FPS:
Architecture:
CNN, YOLO
Instance segmentation identifies objects in an image and maps each pixel to the identified objects. With instance segmentation, you can find exactly where an object is in an image. For example, one could use an instance segmentation model to find all the pixels associated with a forklift in an image.
Instance segmentation models are useful when you need to identify the exact pixels that are connected with an object. This is useful in a number of situations, such as:
The YOLOv5 instance segmentation and the Detectron2 Mask RCNN models are commonly used for instance segmentation.
Roboflow Universe contains over 100,000 open-source datasets and trained models, many of which you can use for instance segmentation tasks. Below are a few of the many models you can use.
This model identifies cracks on concrete. This dataset is usable for those doing transportation and public safety studies, creating self-driving car models, or testing out computer vision models for fun.
This dataset contains 490 annotated images of floors. This model could be used to guide self-driving robots that require knowledge of floors.
Join 100k developers curating high quality datasets and deploying better models with Roboflow.
Get started