Distinctive Image Captioning: Leveraging Ground Truth Captions in CLIP Guided Reinforcement Learning.
Antoine ChaffinEwa KijakVincent ClaveauPublished in: CoRR (2024)
Keyphrases
- ground truth
- reinforcement learning
- image data
- test images
- image features
- multiscale
- single image
- image retrieval
- image classification
- edge detection
- image analysis
- image content
- template matching
- image representation
- image pixels
- low level
- ground truth data
- image structure
- segmented images
- segmentation algorithm
- input image
- machine learning
- function approximation
- web images
- high quality
- keypoints
- markov decision processes
- segmentation method
- region of interest
- image search
- dynamic programming
- image segmentation