About emotionNet to use

Hi guys!

I'm a student interested in facial emotion detection and currently, trying various type of network.

I've used emotionNet by entering "load("models/net_emotion.mat")", and it loaded a emotionNet(which is a DAGNetwork or something... I have no idea).

I tried this one but, this is not that good network cuz whenever I take my selfie as a input, this network classifies it always angry.

So is there any other better emotion classfying network I could get?

Thanks for any recommendation!

Jack

Answer 1

[Aditya Patil]

While you can try out various other networks, another way to improve accuracy is to use transfer learning using a dataset of your images. The get started with transfer learning documentation provides details on how to do this.

For example, you can import a pretrained network, and then fine tune it using transfer learning.