Video Duplicate and Near-Duplicate Detection
Overview
The Near-Duplicate Detection model is used to find duplicate or near-duplicate segments in videos.
Duplicate segments are segments that are identical or nearly identical, even if they have been modified in some way (with overlays, cropping, split-screen, mixing etc.)
Common use-cases include:
- Video Blacklists and Disallow lists: Blacklist videos and prevent them from (re)appearing on your site or app. For instance copyrighted videos, illegal videos, previously removed videos.
- Copyright detection: Detect and manage copyrighted content in user uploads.
- Video set deduplication: Identify and manage duplicate videos within a set.
Image duplicate detection is also available, see the Image Duplicate Detection guide.
What the model catches
Duplicate detection works across all types of videos, both long and short, realistic or animated.
Duplicates are detected across a wide range of transformations and modifications, many of which are typically used to try to evade duplicate detection. Examples:
Original clip
Resolution, size and format changes
- Downscaling and upscaling
- DPI/Resolution changes
- Re-encoding or format conversion (e.g. JPEG, PNG, WEBP...)
Text overlays
- Text overlays and added captions
- Stickers, logos, watermarks
Clip overlays
- Large clip overlays obscuring parts of the original clip
- Emojis, shapes and other graphical overlays
Cropping and reframing
- Tight crops, letterboxing, added borders/frames
- Partial views of the original
Collage
- When the source clip appears inside a multi-clip layout
- Split-screen layouts
Blur
- Strong gaussian blur, motion blur, defocus...
- Pixelation
Clip mixing
- When the source clip is blended with other clips or backgrounds
Color changes and filters
- Grayscale, channel drops, saturation/brightness/contrast/hue changes
- Color filters and tints
Geometric transformations, stretching, flipping
- Aspect ratio changes, stretching, perspective tweaks
- Rotations
- Horizontal and vertical flips
Next steps: learn how to use
Choose your use-case to learn how to use the Video Deduplication model:
Was this page helpful?