AI Video Frame Sampling by Object Movement Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating learning data using video image data is significantly more time and cost-intensive due to its vast data capacity, with preprocessing tasks consuming about 70-80% of the time in AI image learning.
Innovation Solution
A method for sampling a learning target frame image in a video involves extracting frame images at predetermined intervals, detecting and isolating learning target objects, measuring their movement between frames, and selecting frames with sufficient movement to create a learning video set, thereby reducing unnecessary data and preprocessing time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If all video frames are used for learning data generation, then the amount of learning data increases, but the preprocessing time and cost increase significantly
Solution Approach 1:
The video data is segmented into individual frame images at predetermined time intervals, allowing selective processing of only relevant frames rather than processing the entire video continuously. This segmentation enables the system to divide the large video dataset into manageable discrete units for analysis.
Solution Approach 2:
The patent extracts and isolates learning target objects from the video frames by detecting objects with significant movement amounts. By extracting only the frames containing moving objects of interest and removing static backgrounds, the system reduces the data volume while preserving the essential learning content, thereby decreasing preprocessing time without sacrificing data quality.
2Manufacturing precision
If video data is processed in full detail, then the quality of learning data improves, but the processing cost increases by tens to hundreds of times compared to conventional data
Solution Approach 1:
The patent applies local quality by focusing processing resources on specific regions of interest within video frames - namely, the learning target objects that exhibit significant movement. Rather than uniformly processing all pixels and frames with equal detail, the system identifies and processes only the local regions containing moving objects, thereby maintaining high data quality where needed while reducing overall processing complexity and cost.
3Productivity
If frame sampling is performed without object movement analysis, then the preprocessing speed increases, but the quality of learning data decreases
Solution Approach 1:
The patent performs preliminary action by detecting and measuring object movement in each frame before finalizing the learning data selection. By calculating movement amounts of learning target objects between consecutive frames and comparing them against reference values, the system pre-identifies which frames contain meaningful content. This preliminary movement analysis enables rapid filtering of irrelevant frames while ensuring that only frames with significant object movement are selected, thereby maintaining both high preprocessing speed and high data quality.
Data Source
AI summary
Provided is a method, device, and program for sampling a frame image of an object to be learned in a video for artificial intelligence video learning, and an image learning method thereof. The method includes receiving a raw video for the AI image learning, extracting a predetermined number of frame images from the received raw, detecting learning target objects in each of the frame images, removing a background other than the learning target objects from each of the frame images, measuring a movement amount of each of the detected learning target objects in a n-th frame image, from which the background is removed, and selecting the n-th frame image as the learning target frame image, by comparing a result of measuring the movement amount of each of the detected one or more learning target objects in the n-th frame image with a predetermined reference.