Fine Tuning Keras-Facenet Model
47 views
Skip to first unread message
Brihith Shenoy
Mar 14, 2025, 5:29:38 AMMar 14
to Keras-users
So currently I am using a keras-facenet(tf) model for Face Recognition. I use a pretrained model to get the 512-D Embeddings for a person A. Then I give another image as a comparission Image.
Currenttly I have built a vector store and I have plotted all the reference base Images. and then every image is compared against that DB and a conclusion is made.
Would it be possible for me to retrain the model since I have alot of that persons pictures and add like a classification class with that persons name.
Would it be extenedable to more than 1 Class ( 12-15) Cuz I have more than 100 of that persons Images.
Would training be better than the multiple shot classificationg( im not sure if what im doing is 1 shot)
Would the accuracy of classification be better
And How exactly should I do it ( any documentation)
Sorry if the question doesnt make sense
Currenttly I have built a vector store and I have plotted all the reference base Images. and then every image is compared against that DB and a conclusion is made.
Would it be possible for me to retrain the model since I have alot of that persons pictures and add like a classification class with that persons name.
Would it be extenedable to more than 1 Class ( 12-15) Cuz I have more than 100 of that persons Images.
Would training be better than the multiple shot classificationg( im not sure if what im doing is 1 shot)
Would the accuracy of classification be better
And How exactly should I do it ( any documentation)
Sorry if the question doesnt make sense
Reply all
Reply to author
Forward
0 new messages