search

michalfaber / tensorflow_Realtime_Multi-Person_Pose_Estimation

Multi-Person Pose Estimation project for Tensorflow 2.0 with a small and fast model based on MobilenetV3
Other
216 stars 65 forks source link

trained 60 epoch,the validation loss of paf and heatmap is still decrease #3

Open hudengjunai opened 4 years ago

hudengjunai commented 4 years ago

hello,I pull down your code almost no change.I trained the mobilenet based model on 2080Ti with 60 epoch.and I found the val_loss is still decrease 7DA694CA-9D1A-41EB-9E27-8C189367E857 the x-axis is the epoch from 1 to 60,the y-axis is loss.

michalfaber commented 4 years ago

That's good. The lower val loss, the better.

af-yi commented 3 years ago

Excuse me,Are you using the same learning strategy as train_custom_loop_mobilenet.py?