參與:2追蹤:0
人脸识别的算法
針對方案:基于NXP i.mx8m mini的人工智能架构下之人脸辨识解决方案
0得分
最佳解
Dear Sir
Use "MobileFaceNets" on Tensorflow. Converter the pre-trained model to TF Lite mode.
Then use eIQ to build the TF Lite inference enfine.
We can use command:
./FaceRecognition -c 0 -h 0.85” ==> 0.85 is the threshold for the predict score
./FaceRecognition -c 0 -h 0.60” ==> 0.6 is the threshold for the predict score
Thanks
1F