Saliency Map Extraction via Eye Fixation Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for extracting salient objects from images are inefficient due to the time-consuming process of segmenting images into superpixels and processing each one separately, leading to low efficacy.
Innovation Solution
A method and apparatus that utilize a prediction model for conducting convolution, pooling, and normalization processing to obtain eye fixation and semantic description information, which are then fused using an element-wise summation function to generate a saliency map, eliminating the need for image segmentation and reducing processing time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image segmentation into superpixels is performed, then detailed processing of each region can be achieved, but processing time increases significantly
Solution Approach 1:
The patent applies segmentation by dividing the image into superpixels as basic processing units. This allows the model to process smaller, more manageable regions while maintaining overall image context, balancing between detailed processing and computational efficiency
Solution Approach 2:
The patent replaces traditional mechanical image processing methods with a deep learning-based prediction model. The model uses convolutional neural networks to automatically learn features and perform salient object detection, substituting manual or algorithmic segmentation approaches with intelligent, data-driven processing that reduces time loss while maintaining accuracy
2Measurement precision
If each superpixel is processed separately through convolutional neural network, then feature extraction accuracy improves, but overall processing efficiency decreases
Solution Approach 1:
The patent merges the processing of multiple superpixels by using a prediction model that can handle multiple regions simultaneously. The model integrates feature extraction across superpixels through shared convolutional layers and pooling operations, allowing parallel processing that improves efficiency while maintaining accurate feature extraction for each region
Solution Approach 2:
The prediction model employs universal convolutional layers and pooling operations that can process different superpixels using the same learned features. This multi-functional approach allows the model to extract relevant features from various image regions efficiently without requiring separate processing pipelines for each superpixel
Data Source
AI summary
A method and an apparatus for extracting a saliency map are provided in the embodiment of the present application, the method includes: conducting first convolution processing, first pooling processing and normalization processing on an original image via a prediction model to obtain eye fixation information from the original image, where the eye fixation information is used for indicating a region at which human eye gaze; conducting second convolution processing and second pooling processing on the original image via the prediction model to obtain semantic description information from the original image; fusing the eye fixation information and the semantic description information via element-wise summation function; and conducting detection processing on the fused eye fixation information and semantic description information via the prediction model to obtain a saliency map from the original image. It is used for improving the efficiency of extracting the saliency map from image.