Spatial Motion Attention for Noise-Resilient Video Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video analytics systems struggle to accurately focus on moving objects due to noise from background motion and lighting changes in motion images, leading to inefficiencies in tasks like object classification and tracking.
Innovation Solution
Enhance feature maps in a deep learning-based video analytics pipeline using motion images to weight areas with more motion higher and less weight to areas with less motion, generating a motion enhanced feature map for improved object detection and tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion images are used to guide video analytics to focus on moving objects, then the ability to detect moving objects is improved, but noise from background motion and lighting changes degrades the quality of motion images
Solution Approach 1:
The patent introduces an intermediary processing pipeline that takes motion images as input and generates refined motion masks through multiple processing stages including convolutional neural networks, attention mechanisms, and temporal aggregation. This intermediary system filters out noise from background motion and lighting changes while preserving actual object motion, thereby resolving the contradiction between detecting moving objects and eliminating noise.
2Measurement precision
If motion images are used to enhance feature maps, then video analytics accuracy is improved, but the complexity of the processing pipeline increases
Solution Approach 1:
The patent segments the processing pipeline into distinct functional modules: motion image generation, motion mask extraction, feature map enhancement, and analytics processing. Each module performs a specific function and can be independently optimized or configured. This segmentation reduces overall system complexity by making the processing pipeline more modular and manageable while still achieving improved video analytics accuracy.
3Measurement precision
If spatial attention is applied to motion enhanced feature maps, then object detection precision is improved, but computational resources required increase
Solution Approach 1:
The patent applies spatial attention mechanisms that operate locally on regions of the feature maps rather than globally processing entire frames. By focusing computational resources on specific spatial locations where motion is detected, the system achieves improved object detection precision while reducing overall computational resource requirements compared to uniform global processing.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for spatial motion attention for intelligent video analytics. One of the methods includes: obtaining an input image of a region; generating a motion image that characterizes a difference between a value of a pixel at the pixel location in the input image and a value of a pixel at the pixel location in the reference image; generating a feature map using the input image; generating, using the motion image and the feature map, a motion enhanced feature map that has, for one or more pixels that likely indicate movement, a first value that a) indicates that the corresponding pixel in the motion enhanced feature map likely indicates movement and b) is different from a second value for a corresponding pixel in the feature map; and analyzing the motion enhanced feature map.