Compact Neural Network for Real-Time Object Detection on Mobile Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine learning models for object detection in images are typically large and resource-intensive, making them unsuitable for devices with limited processing power and memory, such as smartphones, especially when real-time performance is required for applications like augmented reality.
Innovation Solution
A compact machine-learning model with an optimized architecture that uses convolutional layers and neural processing engines to efficiently perform image-feature recognition tasks, enabling real-time detection and segmentation of objects in images or videos on devices with limited resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional machine learning models are used for object detection, then detection accuracy is improved, but device resource requirements increase and real-time performance cannot be achieved on mobile devices
Solution Approach 1:
The model divides the image processing task into multiple stages: first generating a feature map through convolutional layers, then identifying regions of interest, and finally processing only those regions through additional convolutional layers. This segmentation allows the system to maintain high detection accuracy while reducing overall computational complexity and resource requirements for mobile devices
Solution Approach 2:
Instead of processing the entire image through all convolutional layers, the model applies convolutional layers only to the identified regions of interest after initial feature extraction. This partial action approach reduces processing time and resource consumption while maintaining sufficient accuracy for object detection on mobile platforms
2Measurement precision
If large machine learning models are used, then detection performance is improved, but processing speed decreases and real-time requirements cannot be met
Solution Approach 1:
The model segments the processing pipeline into sequential stages where feature maps are generated first, then regions of interest are identified, and finally only those regions are processed through additional convolutional layers. This segmentation enables the system to achieve real-time processing speeds while maintaining high detection performance by avoiding unnecessary computation on the entire image
Solution Approach 2:
The model dynamically adjusts the processing path based on the identified regions of interest. By using the feature map to guide which regions require further processing, the system can adaptively allocate computational resources, achieving real-time performance without sacrificing detection accuracy
3Measurement precision
If complex machine learning models are deployed on mobile devices, then detection capability is improved, but energy consumption increases
Solution Approach 1:
The model applies convolutional operations only to the identified regions of interest rather than processing the entire image through all layers. This partial action significantly reduces the number of computations required, thereby lowering energy consumption while maintaining sufficient detection capability for mobile applications
Data Source
AI summary
In one embodiment, a system may access an image and generate a feature map for the image using a neural network. The system may identify regions of interest in the feature map. Regional feature maps may be generated for the regions of interest, respectively. Each of the regional feature maps has a first, a second, and a third dimension. The system may generate a first combined regional feature map by combining the regional feature maps. The combined regional feature map has a first, a second, and a third dimension. The system may generate a second combined regional feature map by processing the first combined regional feature map using one or more convolutional layers. The system may generate, for each of the regions of interest, information associated with an object instance based on a portion of the second combined regional feature map associated with that region of interest.