ByteMorph-6M: Image Dataset for Non-Rigid Motion Editing

ByteMorph-6M is a dataset for editing images to reflect non-rigid motions, such as changes in camera viewpoint, object deformation, human articulation, or complex interactions. It was created by ByteDance-Seed to address a gap in current methodologies focused on static or rigid transformations. The dataset was last updated on August 1, 2025.

Use Cases

• Train models for viewpoint synthesis based on the described changes in camera perspective.
• Develop algorithms for object deformation editing based on the described non-rigid motions.
• Research human articulation and pose transfer based on the described complex interactions.
• Benchmark generative models for dynamic scene editing based on the described gap in current methodologies.

Strengths

• Focuses on a significant yet underexplored frontier in computer vision, as stated in the description.
• Addresses a gap by concentrating on non-rigid motions and dynamic movement, unlike datasets for static imagery.
• Last updated on August 1, 2025, indicating recent maintenance.

Limitations

• Row count, file formats, and column-level documentation are unknown, which limits suitability assessment.
• The full description is hosted externally, requiring a visit to the dataset page for complete details.
• License information is unknown, which may restrict usage.

Provenance

Source

ByteDance-Seed on Hugging Face.

Collection Method

The description suggests it was created to bridge a gap in current methodologies, but the specific collection method is not detailed.

Time Range

The temporal coverage of the data itself is unknown.

Freshness

Last updated 2025-08-01 21:21:09.

Geography

The spatial coverage of the data is unknown.