Skip to content

Latest commit

 

History

History
15 lines (8 loc) · 505 Bytes

File metadata and controls

15 lines (8 loc) · 505 Bytes

Home-Surveillance-System-using-Computer-Vision

I have created a face recognizer using convolutional neural networks to recognize the authorized person's faces. I have trained the model on 50 different images of 2 persons using google colab. The accuracy is more than 85%

The CNN model architecture is as follows:

Conv: 16 filters, 3x3 kernel; MaxPooling: 2x2

Conv: 32 filters, 3x3 kernel; MaxPooling: 2x2

Conv: 64 filters, 3x3 kernel; MaxPooling: 2x2

Dense: 512 nodes

Dense: 1 node (output)