TiMix: Text-Aware Image Mixing for Effective Vision-Language Pre-training.
Chaoya JiangWei YeHaiyang XuQinghao YeMing YanJi ZhangShikun ZhangPublished in: AAAI (2024)
Keyphrases
- image data
- image features
- input image
- multiscale
- image content
- single image
- edge detection
- image analysis
- web images
- information retrieval
- image representation
- visual perception
- image classification
- test images
- image pixels
- english text
- low level image processing
- low level
- high resolution
- image collections
- scanned documents
- image segmentation
- complex background
- similarity measure
- region of interest
- training set
- image retrieval
- image matching
- segmentation algorithm
- programming language
- object recognition
- natural language
- segmentation method
- labeled images
- computer vision