ckghosted / disentangle-based-hallucination

disentangle-based data hallucination for few-shot learning
1 stars 0 forks source link

question for AFHN #1

Open HopecColumnSi opened 4 years ago

HopecColumnSi commented 4 years ago

Excuse me,you mentioned it in the script_PN_AFHN_1_tf1_ar1_noBN_noPro.sh about the AFNH. Is the AFNH model in the "Adversarial Feature Hallucination Networks for Few-Shot Learning" paper?

ckghosted commented 4 years ago

Yes, for AFHN I mean the paper "Adversarial Feature Hallucination Networks for Few-Shot Learning" (K. Li et al., CVPR 2020). But I am still working on it. I plan to fix the related codes this week.

HopecColumnSi commented 4 years ago

Do you have the data and source code of the paper "Adversarial Feature Hallucination Networks for Few-Shot Learning" (K. Li et al., CVPR 2020)"? Have you ever implement this paper? Can you share it with me? This is my mail m15666614745@163.com. Thank you very much!

ckghosted commented 4 years ago

I don't have the source code of the AFHN paper. I implemented my own version of AFHN as the HAL_PN_AFHN class in model_hal.py. The datasets used in AFHN (mini-ImageNet, CUB, and CIFAR-100) is available online and the category splits are specified in the data-split folder. For further implementation details, please email me at d05921018@ntu.edu.tw. Thanks!