Detecting device, detecting method and non-transitory computer-readable medium for object detection

Abstract

A detecting device, is used for object detection in an image information processing system. A first computing model of the detecting device detects several objects of different classes in a validation set to generate a detecting result. A merit analysis circuit analyzes the detecting result to generate a highest merit score of each class. A clustering circuit clusters the classes into a first number of clusters based on the highest merit scores, and selects a second number of clusters from the first number of clusters. A normalization calculating circuit calculates an object amount ratio and an average area ratio of each class in the second number of clusters. A global calculating circuit calculates a global confidence threshold based on the object amount ratio and the average area ratio. The first computing model detects objects in several first images of an image metadata of the image information processing system in an inference stage to generate an inference result, and filters the inference result based on the global confidence threshold.

Images