The authors' official implementation of GANalyze, a framework for studying cognitive properties such as memorability, aesthetics, and emotional valence using genenerative models.
Home Page: http://ganalyze.csail.mit.edu/
I'm excited to experiment with this model, but it looks like the memnet weights are broken:
GANalyze/tensorflow$ sh download_pretrained.sh
Downloading AestheticsNet weights
--2019-07-03 11:39:19-- http://ganalyze.csail.mit.edu/models/aestheticsnet_state_dict.p
Resolving ganalyze.csail.mit.edu (ganalyze.csail.mit.edu)... 128.30.100.223
Connecting to ganalyze.csail.mit.edu (ganalyze.csail.mit.edu)|128.30.100.223|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 638536993 (609M) [text/x-pascal]
Saving to: ‘assessors/aestheticsnet_state_dict.p’
aestheticsnet_state_dict.p 100%[===========================================>] 608.96M 86.8MB/s in 7.5s
2019-07-03 11:39:26 (81.0 MB/s) - ‘assessors/aestheticsnet_state_dict.p’ saved [638536993/638536993]
Downloading MemNet weights
--2019-07-03 11:39:26-- http://ganalyze.csail.mit.edu/models/memnet_state_dict.p
Resolving ganalyze.csail.mit.edu (ganalyze.csail.mit.edu)... 128.30.100.223
Connecting to ganalyze.csail.mit.edu (ganalyze.csail.mit.edu)|128.30.100.223|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 511965917 (488M) [text/x-pascal]
Saving to: ‘assessors/memnet_state_dict.p’
memnet_state_dict.p 100%[===========================================>] 488.25M 111MB/s in 5.1s
2019-07-03 11:39:32 (94.9 MB/s) - ‘assessors/memnet_state_dict.p’ saved [511965917/511965917]
--2019-07-03 11:39:32-- http://ganalyze.csail.mit.edu/models/memnet_mean.mat
Resolving ganalyze.csail.mit.edu (ganalyze.csail.mit.edu)... 128.30.100.223
Connecting to ganalyze.csail.mit.edu (ganalyze.csail.mit.edu)|128.30.100.223|:80... connected.
HTTP request sent, awaiting response... 404 Not Found
2019-07-03 11:39:32 ERROR 404: Not Found.
--2019-07-03 11:39:32-- http://ganalyze.csail.mit.edu/models/mean_AADB_regression_warp256.binaryproto
Resolving ganalyze.csail.mit.edu (ganalyze.csail.mit.edu)... 128.30.100.223
Connecting to ganalyze.csail.mit.edu (ganalyze.csail.mit.edu)|128.30.100.223|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 786446 (768K)
Saving to: ‘assessors/mean_AADB_regression_warp256.binaryproto’
mean_AADB_regression_warp25 100%[===========================================>] 768.01K 4.68MB/s in 0.2s
2019-07-03 11:39:32 (4.68 MB/s) - ‘assessors/mean_AADB_regression_warp256.binaryproto’ saved [786446/786446]
--2019-07-03 11:39:32-- http://ganalyze.csail.mit.edu/models/mean_AADB_regression_warp256_lore.npy
Resolving ganalyze.csail.mit.edu (ganalyze.csail.mit.edu)... 128.30.100.223
Connecting to ganalyze.csail.mit.edu (ganalyze.csail.mit.edu)|128.30.100.223|:80... connected.
HTTP request sent, awaiting response... 404 Not Found
2019-07-03 11:39:32 ERROR 404: Not Found.
First of all, thank you for the work.
I wonder why you did not use optimizer.zero_grad() in the torch code ? Is it a technique to get better result or is it just a mistake?
Hi guys,
this is more of an enquiry than an issue. I am interested in working with your framework in the context of memorability, however I can't seem to find a version of MemNet implemented in Pytorch. I only have a CaffeModel of it, but converting it to a Pytorch model has been a struggle. Any input on the matter would be appreciated!
When i install the requirements, i found some requirements conflict. Can you provide me more requirements details? Thanks!
(py36) beryl@beryl-System-Product-Name:~/Documents/github/GANalyze/tensorflow$ conda create -n GANalyze python=3.6 tensorflow-gpu==1.12 numpy scipy PIL
Collecting package metadata (current_repodata.json): done
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: /
Found conflicts! Looking for incompatible packages.
This can take several minutes. Press CTRL-C to abort.
failed
UnsatisfiableError: The following specifications were found to be incompatible with each other:
Package python conflicts for:
python=3.6
numpy -> python[version='2.6.*|2.7.*|3.3.*|3.4.*|3.5.*|3.6.*|>=2.7,<2.8.0a0|>=3.5,<3.6.0a0|>=3.6,<3.7.0a0|>=3.7,<3.8.0a0|>=3.8,<3.9.0a0']
scipy -> python[version='2.6.*|2.7.*|3.3.*|3.4.*|3.5.*|3.6.*|>=2.7,<2.8.0a0|>=3.5,<3.6.0a0|>=3.6,<3.7.0a0|>=3.7,<3.8.0a0|>=3.8,<3.9.0a0']
tensorflow-gpu==1.12 -> tensorflow==1.12.0 -> python[version='2.7.*|3.6.*']
tensorflow-gpu==1.12 -> tensorflow==1.12.0 -> tensorflow-base==1.12.0=gpu_py27had579c0_0 -> mock[version='>=2.0.0'] -> python[version='>=3.6']
tensorflow-gpu==1.12 -> tensorflow==1.12.0 -> tensorflow-base==1.12.0=gpu_py27had579c0_0 -> mock[version='>=2.0.0'] -> pbr -> pip -> wheel -> python[version='2.6.*|3.3.*|3.4.*|3.5.*|>=3.5,<3.6.0a0|>=3.7,<3.8.0a0|>=3.8,<3.9.0a0']
pil -> python[version='2.6.*|2.7.*']
tensorflow-gpu==1.12 -> tensorflow==1.12.0 -> tensorboard[version='>=1.12.0,<1.13.0'] -> python[version='>=2.7,<2.8.0a0|>=3.6,<3.7.0a0']
Package pip conflicts for:
scipy -> python=3.5 -> pip
numpy -> python[version='>=3.8,<3.9.0a0'] -> pip
pil -> python=2.7 -> pip
tensorflow-gpu==1.12 -> tensorflow==1.12.0 -> python=2.7 -> pip
python=3.6 -> pip
I cannot understand how I config the software and doing test
Hi， @LoreGoetschalckx.
I have tried your implementation and it works well.
I have trained my model using these parameters.
python train.py \
--generator biggan256 None \
--assessor emonet \
--transformer OneDirection None \
--train_alpha_a -0.5 \
--train_alpha_b 0.5 \
--gpu_id 0 \
--num_samples 400000 \
--checkpoint_resume 0
But, I have trouble with 3 points.
Given the negative alpha, the model generates only grayscale images.
(here, https://github.com/LoreGoetschalckx/GANalyze/blob/master/pytorch/test_pytorch.py#L267 )
Given the positive alpha, the model generates only one similar sample.
I check the transformed noize in (https://github.com/LoreGoetschalckx/GANalyze/blob/master/pytorch/test_pytorch.py#L152) and notice each transformed z is a similar parameter.
I think I can not learn the model because the hyper parameter is wrong...
Thanks.
when train_tf.py runing , ''state_dict = pickle.load(f, encoding='latin1')'',load memnet_state_dict.p file . A error occurs:_pickle.UnpicklingError: invalid load key, 'H'. Is this memnet_state_dict.p file wrong? Thank you !!!
I keep getting this error log when trying to download memenet_mean.mat for in the "download_pretrained.sh" script in the tensor flow folder. I saw this was a previous issue that was resolved, but the error came back. All the other files downloaded correctly is there anyway to get this file?
--2021-12-14 09:21:01-- http://ganalyze.csail.mit.edu/models/memnet_mean.mat
Resolving ganalyze.csail.mit.edu (ganalyze.csail.mit.edu)... 128.30.100.223
Connecting to ganalyze.csail.mit.edu (ganalyze.csail.mit.edu)|128.30.100.223|:80... connected.
HTTP request sent, awaiting response... 404 Not Found
2021-12-14 09:21:01 ERROR 404: Not Found.
Can anybody solve this problom?
What do I need to realize that? acting in pytoｒｃｈ.
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
Django
The Web framework for perfectionists with deadlines.
Laravel
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft
Open source projects and samples from Microsoft.
Google
Google ❤️ Open Source for everyone.
Alibaba
Alibaba Open Source for everyone
D3
Data-Driven Documents codes.
Tencent
China tencent open source team.