Image Encoding Pipeline for Low-Cost Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods involving neural network models require high costs and inefficient calculations due to the need to process both images and texts, leading to increased computational loads.
Innovation Solution
An image processing method that utilizes a first model to encode and process images independently, determining object locations and categories without requiring text input, employing techniques like multilayer perceptron-based processing, deformable convolution, and attention mechanisms to reduce computational requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the neural network model processes both image and text to perform target detection, then the detection accuracy is improved, but the calculation amount increases and processing efficiency decreases
Solution Approach 1:
The patent extracts and removes the text processing component from the target detection system. Instead of processing both image and text through a neural network model, the invention uses only image encoding and processing to achieve target detection, thereby reducing computational load while maintaining detection capability
Solution Approach 2:
The patent introduces an intermediary approach by using image encoding results as the sole input for target detection processing. The encoding results serve as a mediator that captures essential image features without requiring additional text input, thus simplifying the processing pipeline
2Loss of information
If text input is required for image processing, then the information completeness is improved, but the processing cost increases
Solution Approach 1:
The patent enables the image processing system to be self-sufficient by using only image encoding results for target detection. The system does not require external text input or additional data sources, thereby reducing processing costs while maintaining information completeness through effective image feature extraction
Solution Approach 2:
The patent changes the input parameters of the target detection model from requiring both image and text inputs to accepting only image encoding results. This parameter change reduces the complexity and cost of processing while maintaining the essential information needed for accurate detection
Data Source
AI summary
An image processing method and a related device thereof are provided, to effectively reduce image processing costs, and reduce a calculation amount of image processing, thereby improving image processing efficiency. The method in this application includes: obtaining an image; encoding the image to obtain an encoding result of the image; processing the encoding result to obtain a first processing result of the image, where the first processing result is used to determine locations of M objects in the image, and M≥1; and processing the encoding result and the first processing result to obtain a second processing result of the image, where the second processing result is used to determine categories of the M objects.