深度学习的人体图像半自动标注系统

针对目前数据标注过于依赖硬件、手动数据标注效率低下的问题，提出了基于深度学习的人体图像半自动标注系统.系统通过对算法进行改进，增加人体关键点个数进行特征提取和加入运动信息的约束，提高了视频分阶段标注的准确率.使用真实数据集仿真实验证明了通过深度学习算法进行数据标注的可行性，并且使用半自动标注的速度快、准确率高.

Deep Learning Based Semi-Automatic Labeling System for Human Images

In view of the problem that data labeling is too dependent on hardware and manual data labeling is inefficient,a semi-automatic labeling system for human images based on deep learning is proposed. By improving the algorithm,the system increases the number of key points of the human body for feature extraction and adds motion information constraints,which improves the accuracy of video staged annotation. Experiments that employs real data sets prove the feasibility of data labeling by deep learning algorithm,and using deep learning algorithms for semi-automatic labeling is faster and more accurate.