Labelme vs Roboflow

Labelme is an AI-powered desktop annotation tool for creating image training datasets, aimed at researchers, engineers, and teams. Annotation options include polygons, rectangles, oriented rectangles, circles, lines, points, masks, and rich attributes. Local model support includes SAM, SAM2, SAM3, EfficientSAM, and YOLO-World. SAM3 accepts box and text prompts; SAM2 and EfficientSAM support point prompts, while YOLO-World returns bounding boxes. The maker says images, annotations, and exports stay on the user’s machine, without cloud processing or telemetry. Model weights are downloaded on first use; afterward, annotation and AI assistance can work offline, including in air-gapped setups using transferred models or the offline pack. Pro adds conversion and dataset tools, with exports including YOLO, YOLO-OBB, Pascal VOC, masks, visualizations, and statistics. The application is GPL-3.0, while the Pro toolkit is closed source. It runs on Apple Silicon macOS, 64-bit Windows, and 64-bit Linux; Intel Macs are supported only through v7.0.4. Starter costs $49, paid once, Pro $79, paid once, and Pro (Lifetime) $249, paid once.

Roboflow provides tools for building, deploying, and monitoring computer vision applications, from dataset preparation through model inference. Its dataset features include image annotation, video frame extraction, preprocessing, and augmentation; annotation types include boxes, polygons, segmentation masks, classification labels, and keypoints. Exports include JSON, XML, CSV, TXT, and more than 30 additional formats. A hosted inference API runs models on autoscaling infrastructure with load balancing and burst support. Deployment options include roboflow.js, NVIDIA Jetson, Luxonis OAK, iOS, self-hosting, VPC, and on-premise environments. The Free Tier costs nothing and includes 10 credits monthly, stated as enough for about 30 model trainings or 80,000 inferences. Core costs $39 a month, billed monthly, with 20 credits and a private workspace and models. The pricing comparison lists one model training at a time by default and 20 projects for Core. Roboflow states it encrypts data in transit and at rest and is SOC 2 Type 2 compliant; enterprise pricing is custom.