关于获取unknown object的bounding box的问题

deeplearning-wisc / stud

source code for CVPR'22 paper "Unknown-Aware Object Detection: Learning What You Don’t Know from Videos in the Wild"

Apache License 2.0

117 stars 14 forks source link

Open UMSiyuanLou opened 1 year ago

UMSiyuanLou commented 1 year ago

你好，请问在计算dissimilarity之前，是如何做到在没有见过这个物体的情况下得到它的bounding box的呢？（比如论文图3右图中那些绿色的框）或者说unknown object的feature representation h(x,b) 该如何获取？