A machine learning-based scanning path prediction method and device
A technology of machine learning and prediction methods, applied in the field of image processing, can solve the problems of predicting fixation points relying on static saliency maps, insufficient predicting saccade paths, etc., achieving good universality and scalability, and eliminating dependencies.
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0059] figure 1 It is a schematic flow chart of a method for predicting a glance path based on machine learning in an embodiment of the present invention. Such as figure 1 As shown, the method includes:
[0060] Step 110: Obtain an image data set to be processed, wherein each image information in the image data set has corresponding truth value information;
[0061] Specifically, the image data set to be processed refers to a collection of multiple pictures waiting to be processed, and the corresponding ground truth information refers to the gaze point coordinates of the corresponding image as a label.
[0062] Step 120: making training samples of the image data set according to the true value information;
[0063] Further, the making the training samples of the image data set according to the true value information specifically includes: processing the true value information to obtain eye movement data information of N observers; The observer's eye movement data is subjec...
Embodiment 2
[0092] The effects of the present invention will be further described below in conjunction with simulation experiments.
[0093] 1. Simulation conditions:
[0094] In the simulation experiment of the present invention, the computer system used is Ubuntu 16.04, the machine learning framework is TensorFlow, the version is 1.1.0, the Python version used is 2.7, and the vector of the embedded matrix is V×M, and V is made according to different data sets. Corresponding adjustments, M takes 512, and C takes 16, indicating 8 fixation points.
[0095] 2. Simulation content:
[0096] In the simulation experiment of the present invention, picture names and Arabic numerals are mapped to form a dictionary, an experiment is designed for each data set, training set pictures and test set pictures are selected according to numbers, and corresponding eye movement data sets are processed to obtain labels. Use the samples to train the LSTM network, use the gradient descent optimization algor...
Embodiment 3
[0101] Based on the same inventive concept as a machine learning-based glance path prediction method in the foregoing embodiments, the present invention also provides a machine learning-based glance path prediction device, such as image 3 shown, including:
[0102] A first obtaining unit, the first obtaining unit is used to obtain an image data set to be processed, wherein each image information in the image data set has corresponding truth value information;
[0103] a first production unit, the first production unit is configured to produce training samples of the image data set according to the truth information;
[0104] a second obtaining unit, the second obtaining unit is configured to obtain image feature representation information of the image information according to the image information;
[0105] A first construction unit, the first construction unit is used to construct and train an LSTM network according to the image feature representation information and the ey...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


