基于Faster RCNN的复杂场景目标检测

Study on Target Detection in Complex Scenes Based on Faster RCNN

对于复杂的应用场景,传统的目标检测无法给出满意的检测效果。阐述深度学习方法,采用Faster RCNN目标检测算法实现复杂场景下的目标检测。Faster RCNN目标检测算法借助卷积神经网络、区域提议网络、边界框回归算法,获取目标的区域特征进行目标分类和定位。实验结果表明,将Faster RCNN目标检测算法应用于复杂场景下的目标检测,可显著提高目标的检测效果。

For complex application scenarios, traditional target detection cannot give satisfactory detection results. Based on the deep learning method, Faster RCNN target detection algorithm is used to achieve target detection in complex scenes. The Faster RCNN target detection algorithm uses convolutional neural networks, region proposal networks, and bounding box regression algorithms to obtain the regional features of the target for categorizing the target and determining the location of the target. Experimental results show that applying Faster RCNN target detection algorithm for target detection in complex scenes can significantly improve the target detection effect.