摘要
针对人体胸腰部检测,文章提出一种深度学习的方法对人体胸腰部图像进行目标检测。首先,选用SSD目标检测算法模型(Single Shot MultiBox Detector)并进行微调;其次,利用男性全身图片对该模型进行训练;最后,利用训练完成的模型对人体胸腰部进行识别和定位,并与Mask-RCNN算法模型训练速度和精度进行对比。结果表明,虽然Mask-RCNN算法模型运行速度较快,但SSD目标检测算法相对能够更准确地识别和定位人体胸腰部,准确度达到91.6%,能够有效提高远程在线量身定制中人体胸腰部尺寸检测的准确度。
The paper proposes a deep learning method to perform target detection on human chest and waist images.First,select and fine-tune the SSD(Single Shot MultiBox Detector)target detection algorithm model.Second,use the male body image to train the algorithm model.Finally,use the trained model to identify and locate the human chest and waist.Compared with the training speed and accuracy of the Mask-RCNN algorithm model,results showed that although the Mask-RCNN algorithm model runs faster,the SSD target detection algorithm can relatively accurately identify and locate the human chest and waist,and the detection accuracy reaches 91.6%.This can improve the accuracy of the size of the human chest and waist in the remote online tailor-made.
出处
《纺织导报》
CAS
2020年第11期76-78,共3页
China Textile Leader
基金
上海市科委项目(18030501400)。
