CoCoT: Contrastive Chain-of-Thought Prompting for Large Multimodal Models with Multiple Image Inputs.
Daoan ZhangJunming YangHanjia LyuZijian JinYuan YaoMingkai ChenJiebo LuoPublished in: CoRR (2024)
Keyphrases
- image data
- input image
- image collections
- single image
- image content
- template matching
- image segmentation
- bayesian framework
- image retrieval
- image analysis
- statistical model
- image features
- image classification
- multi modal
- random fields
- image matching
- multiple objects
- multiple models
- keypoints
- segmentation algorithm
- edge detection
- probabilistic model
- high resolution
- low level
- parametric models
- multiscale
- hough transform
- parameter estimation
- image database
- image structure
- pixel values
- image pixels
- surface model
- image statistics