ASSL Method and Apparatus for Open set Object Detection by combining Multi-branch Tree and ASSL

摘要

Provided are a method and a device for detecting an open set object by combining a multi-branch tree and ASSL. The method for detecting an open set object by combining a multi-branch tree and ASSL of the present invention comprises the steps of: training an object detection deep learning model through a multi-branch tree structure by using active semi-supervised learning; and generating subclasses for each object having ambiguity within each class by using information obtained through ambiguity due to a difference in feature information for each object within the same class. Therefore, the object detection deep learning model can be used in various environments with different data features.