Connectionist Network Image Data Processing Offsets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Connectionist networks, such as neural networks, face performance deterioration when classifying spatially transformed objects, especially in outdoor traffic environments with suboptimum image acquisition conditions, due to varying object shapes and forms.
Innovation Solution
A method is introduced that involves determining offsets for individual picture elements in an image, creating a sampling grid to resample input data, which focuses on relevant parts of the image, effectively compensating for shape differences and improving classification accuracy by modifying input feature data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If connectionist networks are trained with standard image data, then classification accuracy is achieved, but performance deteriorates when objects appear in spatially transformed versions with varying shapes
Solution Approach 1:
The patent segments the image processing task by determining individual offsets for each picture element rather than applying a global transformation. This allows different regions of the image to be processed independently, enabling the network to handle varying object shapes and spatial transformations while maintaining classification accuracy.
Solution Approach 2:
The patent applies local quality by determining specific offsets for each picture element based on its position and content. This allows the system to adapt the processing parameters locally for each region, improving the network's ability to recognize objects in spatially transformed versions with varying shapes.
2Device complexity
If the network processes all picture elements uniformly, then processing is simple, but relevant parts of the image are not focused on adequately
Solution Approach 1:
The patent determines different offsets for different picture elements, allowing the system to focus on relevant parts of the image by applying position-dependent processing parameters. This maintains relative simplicity while improving precision in focusing on important image regions.
Data Source
AI summary
A method of processing image data in a connectionist network includes: determining, a plurality of offsets, each offset representing an individual location shift of an underlying one of the plurality of output picture elements, determining, from the plurality of offsets, a grid for sampling from the plurality of input picture elements, wherein the grid comprises a plurality of sampling locations, each sampling location being defined by means of a respective pair of one of the plurality of offsets and the underlying one of the plurality of output picture elements, sampling from the plurality of input picture elements in accordance with the grid, and transmitting, as output data for at least a subsequent one of the plurality of units of the connectionist network, a plurality of sampled picture elements resulting from the sampling, wherein the plurality of sampled picture elements form the plurality of output picture elements.