Multi-adversarial Faster-RCNN with Paradigm Teacher for Unrestricted Object Detection

Multi-adversarial Faster-RCNN with Paradigm Teacher for Unrestricted Object Detection_第1张图片
GRL means ‘gradient reversed layer’,SRM means ‘Scale Reduce Module’.Discriminator submodule at the m m mth block is denoted as D m ^m m
Multi-adversarial Faster-RCNN with Paradigm Teacher for Unrestricted Object Detection_第2张图片
Multi-adversarial Faster-RCNN with Paradigm Teacher for Unrestricted Object Detection_第3张图片

Multi-adversarial Faster-RCNN with Paradigm Teacher for Unrestricted Object Detection_第4张图片
Multi-adversarial Faster-RCNN with Paradigm Teacher for Unrestricted Object Detection_第5张图片
Multi-adversarial Faster-RCNN with Paradigm Teacher for Unrestricted Object Detection_第6张图片
Multi-adversarial Faster-RCNN with Paradigm Teacher for Unrestricted Object Detection_第7张图片
作者未提供代码

你可能感兴趣的:(目标检测,目标检测,人工智能,计算机视觉,算法,深度学习,机器学习,神经网络)