【摘要】 在图像识别与计算机视觉领域,物体检测是研究热点,提出了一种融合多尺度特征的多物体检测方法,基于卷积神经网络在多尺度特征下提取物体的候选区域,然后将不同尺度下的特征进行融合,使多物体检测中出现的小物体被漏检的概率降低.最后采用基于中心点的非极大值抑制方法,计算检测窗口的中心点的欧式距离和iou来抑制冗余的窗口,从而提升多物体检测的精度.将提出的方法在PASCAL VOC数据集上进行验证,实验证明所提的方法能有效提高多物体检测的精确度.更多还原

【Abstract】 In the field of image recognition and computer vision,object detection is a research hotspot.A multi-object detection method that combines multi-scale features is proposed.Based on the convolutional neural network,the candidate regions of the object are extracted under multi-scale features,and then at different scales.The features are fused to reduce the probability of small objects appearing in multi-object detection being missed.Finally,the NMS method based on the center point is used to calculate the Euclidean distance of the center point of the detection window and iou to suppress the redundant window,thereby improving the accuracy of multi-object detection.The method proposed is verified on the PASCAL VOC dataset.The experiment proves that the proposed method can effectively enhance the accuracy of object detection.更多还原