LAION-400M: Open Dataset of CLIP-Filtered 400 Million Image-Text Pairs.
Christoph SchuhmannRichard VencuRomain BeaumontRobert KaczmarczykClayton MullisAarush KattaTheo CoombesJenia JitsevAran KomatsuzakiPublished in: CoRR (2021)
Keyphrases
- image data
- image features
- million images
- image analysis
- input image
- multiscale
- single image
- image dataset
- image content
- image segmentation
- image classification
- low level
- similarity measure
- image representation
- image regions
- high resolution
- test images
- street view
- image pixels
- edge detection
- feature points
- information retrieval
- region of interest
- image set
- image collections
- web images
- weakly labeled
- contrast enhancement
- image search
- semantic information