Tensorflow, Facenet, CV2, MySql - Advanced Face Recognition for Staff Attendance & Movements
3.8 هزار بار بازدید -
5 سال پیش
-
Subscribe Now@whitedigitalnagercoilThis
Subscribe Now
@whitedigitalnagercoil
This is a revised version of POC on how we can use the real-time face recognition by facenet (tensorflow) to detect and identify known faces from a Camera. We are building a virtual attendance system based on Python, Tensorflow, Facenet & CV2. Thanks to AISangam who was our inspiration Real-time face recognition on custom ... and to David Sandberg of FACENET.
https://github.com/davidsandberg/facenet
The images (faces) of staff were trained with an average of 15 photos per staff. We trained the facenet model with these images after data augmentation (Approx. 40 images/class). The system detects the faces, draws a bounding box if the face size is over 26x26 pix and greets the staff. They are checked for a probability over .73 and rest of the faces are considered unknown. We were able to detect 98% of the faces which are recorded in a MySQL DB along with time/late time etc.
https://www.whitedigital.in is an award-winning Google Premier Partner Agency and ranked as the “#1 PPC Agency” in India by Silicon India. As a Premier Google Partner Agency, we employ Google certified engineers to create, manage and optimize Google Ads campaigns that consistently drive Max Return on Investment ( ROI ) results for our clients. Recently we have ventured into AI & Machine learning. We are working on several other interesting projects including chatbot etc. Please feel free to contact [email protected] for any guidance.
@whitedigitalnagercoil
This is a revised version of POC on how we can use the real-time face recognition by facenet (tensorflow) to detect and identify known faces from a Camera. We are building a virtual attendance system based on Python, Tensorflow, Facenet & CV2. Thanks to AISangam who was our inspiration Real-time face recognition on custom ... and to David Sandberg of FACENET.
https://github.com/davidsandberg/facenet
The images (faces) of staff were trained with an average of 15 photos per staff. We trained the facenet model with these images after data augmentation (Approx. 40 images/class). The system detects the faces, draws a bounding box if the face size is over 26x26 pix and greets the staff. They are checked for a probability over .73 and rest of the faces are considered unknown. We were able to detect 98% of the faces which are recorded in a MySQL DB along with time/late time etc.
https://www.whitedigital.in is an award-winning Google Premier Partner Agency and ranked as the “#1 PPC Agency” in India by Silicon India. As a Premier Google Partner Agency, we employ Google certified engineers to create, manage and optimize Google Ads campaigns that consistently drive Max Return on Investment ( ROI ) results for our clients. Recently we have ventured into AI & Machine learning. We are working on several other interesting projects including chatbot etc. Please feel free to contact [email protected] for any guidance.
5 سال پیش
در تاریخ 1397/12/22 منتشر شده
است.
3,815
بـار بازدید شده