基于回归与深度强化学习的目标检测算法

Deep Reinforcement Learning for Object Detection with Regression

基于强化学习的目标检测算法在检测过程中通常采用预定义搜索行为,其产生的候选区域形状和尺寸变化单一,导致目标检测精确度较低。为此,在基于深度强化学习的视觉目标检测算法基础上,提出联合回归与深度强化学习的目标检测算法。首先,深度强化学习agent根据初始候选区域所提取的信息决定相应搜索行动,根据行动选择下一个逼近真实目标的候选区域;然后,重复上述过程,直至agent能确定当前区域为目标区域时终止搜索过程;最后,由回归网络对当前区域坐标进行回归,达到精确定位目的。实验结果显示,在单类别目标检测中,与原算法相比其精度提高了5.4%,表明通过引入回归有效提高了目标检测精确度。

The object detection algorithm based on reinforcement learning usually adopts predefined search actions in the detection process and the shape and size of the proposal regions generated by them are not changed much,resulting in low accuracy of object detection.For this reason,based on the deep reinforcement object detection algorithm,we proposed an object detection algorithm by combining regression with deep reinforcement learning.Firstly,the agent determines the search action according to the information extracted from the initial proposal regions,and then selects the next proposal region approaching the ground truth according to the action.Then the above process is repeated until agent has enough confidence to determine the current region as the ground truth,and then the search process is terminated.Finally,the current region coordinates are regressed by the regression network to achieve a better localization.Compared with the original algorithm,the accuracy of single-class object detection is improved by5.4%,which indicates that the accuracy of visual object detection is improved effectively by introducing regression.