Parallel Neural Network Paths for Image Processing Latency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing computer vision systems face inefficiencies in processing image data to detect objects and their characteristics simultaneously, leading to increased latency and reduced processing speed, especially when dealing with real-time video streams.
Innovation Solution
A neural network architecture with parallel paths processes image data to detect objects and their characteristics in parallel, sharing initial layers and using scale-specific groups of layers for efficient detection and recognition, allowing for simultaneous object and characteristic identification, reducing latency and improving processing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If sequential processing of objects and characteristics is used, then processing accuracy is maintained, but processing speed decreases and latency increases
Solution Approach 1:
The neural network is segmented into multiple parallel paths (first path for object detection, second path for characteristic detection) that share initial layers. This segmentation allows simultaneous processing of objects and characteristics without sequential delays, resolving the contradiction between processing speed and latency.
Solution Approach 2:
The invention transitions from sequential (one-dimensional) processing to parallel (multi-dimensional) processing by creating separate processing paths for objects and characteristics. This dimensional change enables both tasks to occur simultaneously, improving speed while reducing latency.
2Productivity
If parallel processing paths are used, then processing efficiency increases, but device complexity increases
Solution Approach 1:
The first and second processing paths share initial layers of the neural network, merging common processing functions into a single set of layers. This reduces the overall complexity of the parallel architecture while maintaining the productivity benefits of simultaneous object and characteristic detection.
Solution Approach 2:
The shared initial layers serve multiple functions by processing features for both object detection and characteristic detection simultaneously. This multi-functionality reduces redundancy and simplifies the overall device complexity while maintaining high processing efficiency.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
There is described a computer-implemented system for processing image data, the system comprising: a sensor operable to capture image data comprising an image of an environment of the sensor; and a processing circuit comprising: a processor; and a computer-readable storage medium comprising computer-readable instructions which, when executed, cause the processor to: receive the image data from the sensor; process the image data using a first path through a neural network to obtain first data, the first path being configured to indicate a presence in the environment of one or more objects of a predetermined object type; process the image data using a second path through a neural network to obtain second data, the second path being configured to indicate a presence in the environment of one or more object characteristics corresponding to the predetermined object type; and generate output data using the first and the second data, wherein the first path and the second path are arranged to be enable the first data and the second data to be obtained in parallel.