WEA-DINO: An Improved DINO With Word Embedding Alignment for Remote Scene Zero-Shot Object Detection.
Guangbiao WangHongbo ZhaoQing ChangShuchang LyuGuangliang ChengHuojin ChenPublished in: IEEE Geosci. Remote. Sens. Lett. (2024)
Keyphrases
- object detection
- scene understanding
- scene recognition
- object categories
- word level
- live video
- object recognition
- object detectors
- word alignment
- d scene
- face detection
- single image
- scene categorization
- scene classification
- three dimensional
- video sequences
- real scenes
- object hypotheses
- moving objects
- multiple images
- vision system
- co occurrence
- true multi image
- human detection
- ground plane
- object class
- dynamic scenes
- image sequences
- dynamic time warping
- pedestrian detection
- sentence level
- word segmentation
- word recognition
- background subtraction
- input image
- image features
- computer vision
- sentence pairs
- real time