I know that the facial recognition in the example NNFaceDetectionRecognition stores the feature values in flash memory, but today I want to use multiple AMB82-mini devices to create an edge access control system. My current idea is to first change the storage of the feature values from facial recognition results to an SD card. However, the problem I am facing now is that I can’t find the variable for the ‘feature values’ or ‘identified facial data.’ Does anyone have a solution or relevant experience with this issue?
Hi @JohnnyKo, May I know what do you mean by feature values? Do you mean the coordinates of each recognized face or the facial points such as eyes, nose and mouth?
You can study the API documents:
https://www.amebaiot.com/en/rtl8735b-arduino-api-neuralnetwork-facedetection/
https://www.amebaiot.com/en/rtl8735b-arduino-api-neuralnetwork-facerecognition/
The results of detected faces are stored in a vector, and xMin, xMax, yMin, yMax are the coordinates of the bounding box containing the recognized face. if you are looking at getting facial points such as eyes, nose, and mouth. You can consider referring to NNFaceDetection. xFeature and yFeatures APIs are used to identify the facial data.
There’s an example here that will capture an image and store to SD card: Neural Network – Capture JPEG Face Recognition – Realtek IoT/Wi-Fi MCU Solutions. Not sure if its something you are looking for.
Hi @pammyleong ,
I want to share the features extracted from the pretrained model MobileFaceNet with each of my AMB82-mini devices located in different places. My goal is to record facial features on AMB82-mini-1, and then directly perform facial recognition on AMB82-mini-2 without having to extract and record facial features again on AMB82-mini-2.
I am considering the method of capturing an image and storing it to an SD card using the Neural Network – Capture JPEG Face Recognition – Realtek IoT/Wi-Fi MCU. However, this approach might not resolve my current issue. Could you suggest any other solutions that might help with my problem? Thank you!