Target detection method and device based on yolk network, and equipment terminal

Abstract

The invention relates to a target detection method and device based on a yolo network, an equipment terminal and a readable storage medium, and the method comprises the steps: obtaining the initial preset feature complexity of each target type in training data, and obtaining the initial preset feature complexity of each target type based on the yolo network according to the initial preset feature complexity of each target type and a plurality of preset complexity intervals; establishing a respective feature layer structure of each preset complexity interval, calculating and adjusting a loss function value of each preset complexity interval based on the respective feature layer structure and the initial preset feature complexity of each target category, and carrying out the splicing of loss functions according to the loss function value of each preset complexity interval, so as tocalculate a loss value of the spliced loss function; performing weight and offset updating according to the loss value to generate a target detection model, testing test data according to the target detection model, and outputting position information and category information of a target, thereby improving the accuracy of target detection.

Description

technical field [0001] The present application relates to the technical field of target detection, and in particular to a target detection method, device, device terminal and readable storage medium based on a yolo network. Background technique [0002] At present, the yolo network (You only look once, you only look once) is widely used in target detection technology. However, for the same target data set, there are often multiple category features with different category feature richness. Therefore, if conventional training is used The method cannot effectively extract the key features of the target category, which makes the yolo network difficult to converge. SUMMARY OF THE INVENTION [0003] In view of this, the present application provides a target detection method, device, device terminal and readable storage medium based on yolo network, so as to solve the problem that the existing yolo network detects the same target data set containing different category features a...

AI Technical Summary

Problems solved by technology

[0003] In view of this, the present application provides a target detection method, device, device terminal and readable storage medium based on the yolo network to solve the problem that the existing yolo network detects the same target data set with different richness of category features. technical problems that cannot be converged

Images