When a user upload a selfie, the model search same person in dataset of images of multiple persons and get back all the images in which that person is present.

Step 1: From dataset of images I detect all the faces and save the cropped faces by DeepFace(RetinaFace) and OpenCV.

Step 2: Read all the Cropped faces by OpenCV, resize it, store into a numpy array and standardize it.

Step 3: Feed it to VGG16(Keras) model without including top layer with 0 trainable parameters(use the weights) and extract the features from image numpy array.

Step 4: Feed these extracted features to clustering model(unsupervised)

I tried KMeans and Agglomerative Hierarchical Clustering but not getting as expected.

The clustering is not accurate at all, way more out of the way.

Can anyone help me to address this, how to approach this task, in which step I am making mistakes...

The images/data is not labeled and want to build unsupervised model of clustering.

Any help or suggestions will help me a lot.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.