aimerykong / OpenGAN

ICCV2021 - training a post-hoc lightweight GAN-discriminator for open-set recognition
122 stars 17 forks source link

utils import #2

Open Symbiomancer opened 3 years ago

Symbiomancer commented 3 years ago

Is there a "utils" folder that wasn't uploaded? There are a few imports for utils.*

Thanks

aimerykong commented 3 years ago

Hi Matthew,

I uploaded the utils folder just now. I will take another pass to make sure the naming of the functions is consistent with what is used by the jupyter notebook files. But all functions should be in utils now.

Regards, Shu

On Thu, Sep 16, 2021 at 3:13 PM Matthew Evanusa @.***> wrote:

Is there a "utils" folder that wasn't uploaded? There are a few imports for utils.*

Thanks

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/aimerykong/OpenGAN/issues/2, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABRJSJB37LAO2WKMSWF6U4LUCI6T7ANCNFSM5EFNXHWQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

Symbiomancer commented 3 years ago

Much appreciated! Thank you so much.

Matthew Evanusa PhD Candidate, Computer Science UMD College Park Conductor Emeritus, UMD GSO

On Thu, Sep 16, 2021, 9:31 PM Shu Kong @.***> wrote:

Hi Matthew,

I uploaded the utils folder just now. I will take another pass to make sure the naming of the functions is consistent with what is used by the jupyter notebook files. But all functions should be in utils now.

Regards, Shu

On Thu, Sep 16, 2021 at 3:13 PM Matthew Evanusa @.***> wrote:

Is there a "utils" folder that wasn't uploaded? There are a few imports for utils.*

Thanks

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/aimerykong/OpenGAN/issues/2, or unsubscribe < https://github.com/notifications/unsubscribe-auth/ABRJSJB37LAO2WKMSWF6U4LUCI6T7ANCNFSM5EFNXHWQ

. Triage notifications on the go with GitHub Mobile for iOS < https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675

or Android < https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub .

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/aimerykong/OpenGAN/issues/2#issuecomment-921379705, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADUYUXTCTIUCNATCGPUFZ23UCKK7VANCNFSM5EFNXHWQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

alexander-g commented 3 years ago

Hi @aimerykong, could you please also provide the feature files (feats/res18sc.pkl) and how they were generated? Were they extracted from a classifier trained on the target dataset (cifar/svhn) or on ImageNet?

aimerykong commented 3 years ago

Hi Alex, the features are extracted from the off-the-shelf models, which were trained from-scratch on the closed-set dataset only (hence the mark "sc" in "res18sc" short for "scratch"). That said, we did not use ImageNet to pretrain the models. For example, for CIFAR10, I train a 10-way classification network from scratch using Res18 architecture, and extract the features at the penultimate layers (pre-logit layer). Actually, our offline work shows that better performance can be achieved by ImageNet-pretraining and combining features from different layers. We did not report this in the paper just to keep the paper "simpler" and aligned with the literature.

On Fri, Sep 24, 2021 at 6:15 AM alexander-g @.***> wrote:

Hi @aimerykong https://github.com/aimerykong, could you please also provide the feature files (feats/res18sc.pkl) and how they were generated? Were they extracted from a classifier trained on the target dataset (cifar/svhn) or on ImageNet?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/aimerykong/OpenGAN/issues/2#issuecomment-926512529, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABRJSJD6MGU4OIQ5ZZFN533UDRFS7ANCNFSM5EFNXHWQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

KyleWang-Hunter commented 3 years ago

could you please also provide the feature files, i just want to rerun the result。i am a new fish ,thank you very much。