Bounding boxes, polygons, segmentation, keypoints, bitmasks.
Data Type Coverage
How each platform handles your data
Taskmonk
Encord
Image
STRONG
Bounding boxes, polygons, segmentation, keypoints, bitmasks
STRONG
Same + SAM3 auto-segment, micro-model pre-labeling
Video
STRONG
Frame sync, object tracking, sensor fusion for ADAS and warehouse robotics
STRONG
Native rendering, object tracking, a core product strength. No downsampling.
Audio
STRONG
Transcription, diarization, emotion tagging, multimodal audio-visual alignment
MATURING
Added late 2024. Available, still building depth.
Text / NLP
STRONG
NER, classification, sentiment, document automation, multilingual / vernacular
STRONG
Text, HTML, documents. RLHF and LLM fine-tuning workflows added in 2024.
LiDAR / 3D
STRONG
Sensor fusion, calibrated extrinsics, time-sync, overlaid multi-sensor views
LIMITED
Supported but not a primary use case. Physical AI roadmap still maturing.
DICOM
STRONG
75,000+ DICOM studies annotated. 98.7% diagnostic AI accuracy. HIPAA + ISO 27001 compliant.
STRONG
Purpose-built tool. 20,000+ pixel intensities. Named clients: Mayo Clinic, Stanford Medicine, Northwell Health.
Geospatial
STRONG
Satellite imaging, ADAS, and aerial workflows. Dedicated geospatial annotation page.
NOT DOCUMENTD
No dedicated geospatial workflow on site.




