Improving the age accuracy for Age and Gender Detection Deep learning project

Above is project file of age and gender prediction model and i am having problem with age prediction, its not predicting age accurately. if someone could help me in solving issue Asap!

Hi @samir_khan - are you having issues with both gender and age?

Gender should be fairly easy; I would just have an object detection model (or classification, if each image only contains one person) that only detects male and female.

For age, this is a bit trickier with the models available today. Generally I suggest ranging the model (so one class might be 0-15 years old, one class is 16-30, one class is 31-50, one class is 50+). You’ll need to pick ranges that fit the criteria of your project, though I’ll warn that getting a discreet estimate will be hard.

Yes i am aware of this but I really want it to at least predict 30% accurate age, is their any way you guys could help me with this?

How is your class ontology set up now? Do you have a different class for each age (e.g., 15, 16, 17`)?

No, I haven’t used any classes into separate groups, just 0 to 100+ (e.g., 15 , 16 , 17`)
yes, I have used a different class for each age
in this project file: Age and Gender application for farah.rar - Google Drive

I suggest starting by using groups for your age - the fewer classes to start, the better

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.