Video Similarity Matching via Static Region Cropping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video information processing technologies struggle with low accuracy in determining video similarity due to the difficulty in resolving semantic understanding of images, which is essential for integrating computer vision and natural language processing.
Innovation Solution
A method involving determining a video image frame set, identifying static stitching regions, cropping image frames based on these regions, and using a video information processing model to calculate image feature vectors for similarity comparison between videos.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional video information processing technology is used to determine video similarity, then the processing can be performed on large amounts of video information, but the accuracy of similarity recognition is low
Solution Approach 1:
The patent segments video frames into multiple regions (foreground, background, static, dynamic regions) and processes each region separately to extract relevant features. This segmentation approach enables accurate similarity recognition by focusing on meaningful content regions while ignoring irrelevant areas, thereby improving measurement precision without excessive complexity increase
Solution Approach 2:
The patent extracts key features from specific video regions (such as foreground objects and dynamic regions) while discarding irrelevant information like static backgrounds. This extraction principle improves accuracy by concentrating computational resources on the most discriminative features that actually contribute to video similarity determination
2Measurement precision
If semantic understanding of images is implemented to improve video similarity recognition, then the accuracy improves, but the computational complexity and processing difficulty increase
Solution Approach 1:
The patent introduces region classification as an intermediary step between raw video frames and similarity computation. By classifying regions into foreground/background and static/dynamic categories, the system creates intermediate representations that capture semantic meaning without requiring full semantic understanding, thus improving accuracy while managing complexity
Solution Approach 2:
The patent applies different processing strategies to different regions of video frames based on their semantic importance. Foreground and dynamic regions receive more detailed analysis while static backgrounds are processed differently or excluded, allowing the system to achieve good semantic understanding where needed without uniformly high complexity across the entire frame
Data Source
AI summary
This application provides a video information processing method performed by an electronic device. The method includes: determining a video image frame set corresponding to each of a first video and a second video, respectively; determining a static stitching region corresponding to image frames in the video image frame set; cropping the image frames in the video image frame set according to the static stitching region, and determining an image feature vector for the video based on a corresponding cropping result using a video information processing model; and determining a similarity between the first video and the second video based on an image feature vector corresponding to the first video and an image feature vector corresponding to the second video.