We will use The IMDB-WIKI dataset, and extract face before pushing it into our DL model, the Rasmus Rothe's paper use vgg16, so it will be slow if you want to run real-time, I can use SquezzeNet
Relevant Skills and Experience
- Build many DL models: image classification, face detection, image caption
- Build small memory model so you can run realtime one cheap device such mobile phone or CPU
- Tensorflow, Keras, Caffe
Proposed Milestones
$50 USD - extract face, data augmentation
$55 USD - build Rasmus Rothe's model
$50 USD - small memory and less computation for realtime
Additional Services Offered
$5 USD - API if you want this model for production.
Do you want to run it locally on device or CPU server?