End-to-end target tracking method based on hierarchical feature representation
A target tracking and feature extraction technology, applied in the field of deep neural network, can solve the problems of large amount of calculation and time-consuming, and achieve the effect of improving the effect
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment Construction
[0045] The detailed parameters of the present invention will be further specifically described below.
[0046] Such as figure 1 As shown, the present invention provides a deep neural network framework for target tracking.
[0047] Step (1), data preprocessing, feature extraction
[0048] For the image pair (x', y'), where x' is the template image frame, the template image frame x' is preprocessed and scaled to a size of 127*127; y' is the search image frame, and the search image frame y' is processed The preprocessing is scaled to a size of 255*255; then a network flow of the Siamese network is used to calculate their respective feature representations. Here we use ImageNet's video target detection data set as training data and OTB-100 as test data. For image pair data, the existing Alexnet network model is used to extract image features. Specifically, the template image in the image data pair is scaled to 127×127, the search image size is scaled to 255×255, and input into...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More 


