about test result

[jolinlinlin]

Thank you for public your model and code. Do you test your model in AFLW2000? How about the test result? I follow the test code in https://github.com/natanielruiz/deep-head-pose, and I get a not good result by shuff_epoch_120.pkl:Yaw: 19.8251, Pitch: 9.0400, Roll: 8.1506. I don't know why. During testing, I replace x = x.mean([2, 3])to x = x.mean(3).mean(2) in stable_hopenetlite.py because my torch version is 0.4.1. I don't know whether it is the reason for bad result. Hope to see your test result. Thank you very much!

[stevenyangyj]

Thank you for public your model and code. Do you test your model in AFLW2000? How about the test result? I follow the test code in https://github.com/natanielruiz/deep-head-pose, and I get a not good result by shuff_epoch_120.pkl:Yaw: 19.8251, Pitch: 9.0400, Roll: 8.1506. I don't know why. During testing, I replace x = x.mean([2, 3])to x = x.mean(3).mean(2) in stable_hopenetlite.py because my torch version is 0.4.1. I don't know whether it is the reason for bad result. Hope to see your test result. Thank you very much!

Hi, thanks for ur question. As I said, the model is light and takes a trade-off between speed and accuracy. So it is impossible if you want a free lunch :-). I just trained this model in the 300W-LP dataset and did not take any data augmentation strategy. Sorry, I cannot tell you whether your test result is correct because I have not tested this model in AFLW2000 yet. I just released a toy or demo model (Due to the knowledge protection policy of company, I won't release more powerful model, please understand). If you want to get a model for industry production or paper research, I recommend you to re-train this light model with careful hyper-parameter tuning and data augmentation strategy.

Thanks