How do I annotate a dataset in FiftyOne? | Voxel51

How do I annotate a dataset in FiftyOne? Eric Hofesmann, founding ML engineer and product manager at Voxel51, annotates a malaria cell image dataset step by step in the FiftyOne App: define a label schema, add classification and segmentation fields, then annotate with brush, polygon, and SAM 2 click-to-segment. You'll see how to configure a label schema in the Annotate tab, including a classification field with parasitized and uninfected classes and a detection field for segmentation masks with an occlusion attribute on a 0 to 1 slider. Then Eric annotates a sample three ways: brush, polygon pen, and AI annotation with SAM 2, refining the mask with positive and negative keypoints. SAM 2 tiny runs in the browser in open source. GPU-backed SAM 2 large runs on Voxel51. This is the third video in FiftyOne Questions, a series where each video answers one question about FiftyOne and Voxel51, our enterprise offering. 📝 Read the full post, with all three videos: voxel51.com/blog/in-app-annotation-fiftyone ⏱️ Chapters 0:00 The dataset and the goal: detecting malaria parasites 1:01 Step 1: Open the Annotate tab and configure a label schema 2:28 Create a classification field and add classes 3:16 Create a detection field for segmentation with an occlusion attribute 4:48 Step 2: Annotate the image-level classification 5:14 Segmentation with the brush tool 5:57 Segmentation with the polygon pen 6:17 AI annotation with SAM 2: tiny in-browser vs. large on GPU 7:08 Refine the mask with positive and negative keypoints 8:24 Step 3: Repeat across samples 9:05 27,000 samples: where annotation workflows come in ▶️ FiftyOne Questions How does visualization work in FiftyOne? https://youtu.be/E6xJGXIhpOM What can I annotate in FiftyOne? https://youtu.be/WrrThA30S5I Next: annotation workflows at scale [coming soon] 🔗 Learn more In-app annotation docs: https://docs.voxel51.com/user_guide/annotation.html What is a label schema? https://voxel51.com/glossary/label-schema What is click-to-segment? https://voxel51.com/glossary/click-to-segment SAM 2 (Segment Anything Model 2): https://voxel51.com/glossary/sam-2-segment-anything-model-2 FiftyOne open source: https://voxel51.com/fiftyone Voxel51 for annotation at scale: https://voxel51.com/annotation Get started: pip install fiftyone 💬 Join the community: https://community.voxel51.com FiftyOne is the open source multimodal data platform for physical AI. Voxel51 is the enterprise platform built on it, adding annotation workflows, auto labeling, GPU-backed models, and team collaboration. Together they help ML teams curate, annotate, and evaluate data for visual AI and physical AI models, across images, video, 3D, and multimodal sensor data. #FiftyOne #DataAnnotation #SAM2