How can I get a more stable model？

yu4u / age-gender-estimation

Keras implementation of a CNN network for age and gender estimation

MIT License

1.47k stars 503 forks source link

How can I get a more stable model？ #60

Open tirtile opened 6 years ago

tirtile commented 6 years ago

Thank you for sharing, I have a question. When using cam to test, the face is almost unchanged, but the probability of the model output is very unstable. Do you have any good solutions?

yu4u commented 6 years ago

One easy way is test-time augmentation; augment test each of face images, do prediction against augmented images, and average the prediction results. (though I have not tested...)