Supervisely

3D bounding box annotation with automatic adjustment and tracking

Advanced alpha channel matting for image annotations

Multi-planar and 3D perspective annotation for medical imagery with HIPAA compliance and anonymization

Support for 64-bit and higher color depth images

Support for huge images with 1000+ labels

Best-in-class labeling tools for creating high-quality training data across various image types

Custom metadata tagging for annotations

Skeleton and keypoint labeling tools

Comprehensive labeling of 3D scenes from LiDAR or RADAR sensors with photo and video context

Support for multi-camera and multi-stream video annotation

Support for multi-spectral images beyond standard RGB

Annotation tools for multi-view image datasets

Support for complex hierarchical label structures

Tools for detecting objects in images and video

Tracking annotations across video frames with AI assistance

Pixel-level and instance segmentation annotation tools

Synchronize 2D and 3D annotations with multi-modal sensor data

Support for ultra-wide angle and panoramic images

Annotate hours-long videos with auto-tracking and off-screen detection without cutting into images

Advanced object tracking with support for multiple tracking algorithms

Create custom data augmentation pipelines

Organize and manage datasets with advanced search, visualization and filtering

Support for hundreds of popular formats including PascalVOC, COCO, and custom formats

Tools to migrate from Labelbox, Roboflow, V7 and other platforms

Tools for data augmentation and pre-processing

Natural language search and advanced filtering with embeddings plots visualization

Generate synthetic training data to augment datasets

Human-in-the-loop framework to improve models iteratively

Automated end-to-end training pipelines

Export model checkpoints without vendor lock-in

Train models on cloud infrastructure without local resources

Integration of custom trained models into the platform

Access to foundation models for transfer learning and pre-training

Tools to experiment, manage and compare model versions with performance metrics

Infrastructure to train state-of-the-art neural networks on datasets

Collection of pretrained models for immediate use

Comprehensive activity logs for audit and tracking

Support for custom SSL certificates for enterprise deployments

Tools for anonymizing sensitive medical data

HIPAA-compliant features for medical data handling

Platform deployment on secure private networks without internet access

Fine-grained permissions and custom roles for different user types

OpenID, LDAP, OAuth2 and Microsoft Active Directory support for enterprise SSO