How to put the data to evaluate the result?

[cookie-ke]

Could you please tell me how to put the image data into a folder when I want to use FID, LPIPS, or IS? The train set and data set are split by ".pikle" files and I don't know how to the evaluation metric.

[wtliao]

Could you please tell me how to put the image data into a folder when I want to use FID, LPIPS, or IS? The train set and data set are split by ".pikle" files and I don't know how to the evaluation metric.

Sorry for so late reply. Hope you have solved it. But I also give an example here in case not.

• In IS.py, you need to set argument --input_image_dir to point to the folder of your generated images.
• In test_lpips.py, you need to set --orig_image_path to point to the original image. Note that, you need to resize the original image to the same size as your generated image. Then, you need to set --generated_image_path to point to your generated image, as in IS.py.
• For FID evaluation, please refer to https://github.com/bioinf-jku/TTUR. I implemented their code.

[priyankaupadhyay090]

Could you please tell me how to put the image data into a folder when I want to use FID, LPIPS, or IS? The train set and data set are split by ".pikle" files and I don't know how to the evaluation metric.

Hey, were you able to solve this? I want some help to generate images?

[cookie-ke]

Could you please tell me how to put the image data into a folder when I want to use FID, LPIPS, or IS? The train set and data set are split by ".pikle" files and I don't know how to the evaluation metric.

Hey, were you able to solve this? I want some help to generate images?

I just generated the original images during the test stage so I could compare them with fake image to obtain the metric result.

[priyankaupadhyay090]

Could you please tell me how to put the image data into a folder when I want to use FID, LPIPS, or IS? The train set and data set are split by ".pikle" files and I don't know how to the evaluation metric.

Hey, were you able to solve this? I want some help to generate images?

I just generated the original images during the test stage so I could compare them with fake image to obtain the metric result.

before generating the images. I have a doubt into the sampling() method of main.py. Can I contact you on email ?

[cookie-ke]

Could you please tell me how to put the image data into a folder when I want to use FID, LPIPS, or IS? The train set and data set are split by ".pikle" files and I don't know how to the evaluation metric.

Hey, were you able to solve this? I want some help to generate images?

I just generated the original images during the test stage so I could compare them with fake image to obtain the metric result.

before generating the images. I have a doubt into the sampling() method of main.py. Can I contact you on email ?

You can contact me through GitHub if I can help you. I am now on vacation at home and my experiment codes are in the school laboratory.

[priyankaupadhyay090]

so when I run main.py. I get this

Total filenames: 11788 001.Black_footed_Albatross/Black_Footed_Albatross_0046_18.jpg
Load filenames from: data/birds/train/filenames.pickle (8855)
Load filenames from: data/birds/test/filenames.pickle (2933)
Load from: data/birds/captions.pickle
5450 10
/opt/conda/lib/python3.6/site-packages/torch/nn/modules/rnn.py:61: UserWarning: dropout option adds dropout after all but last recurrent layer, so non-zero dropout expects num_layers greater than 1, but got dropout=0.5 and num_layers=1
"num_layers={}".format(dropout, num_layers))
Downloading: "https://download.pytorch.org/models/inception_v3_google-1a9a5a14.pth" to /root/.cache/torch/hub/checkpoints/inception_v3_google-1a9a5a14.pth
100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 104M/104M [00:01<00:00, 64.4MB/s]
Load pretrained inception v3 model
Load image encoder from: DAMSMencoders/bird/image_encoder200.pth
Traceback (most recent call last):
File "main.py", line 547, in
sampling(text_encoder, netG, dataloader, ixtoword, device) # generate images for the whole valid dataset
File "main.py", line 76, in sampling
start_epoch = int(cfg.TRAIN.NET_G[istart:iend])
ValueError: invalid literal for int() with base 10: ''

this is the main.py
def sampling(text_encoder, netG, dataloader, ixtoword, device):
model_dir = cfg.TRAIN.NET_G
istart = cfg.TRAIN.NET_G.rfind('_') + 1
iend = cfg.TRAIN.NET_G.rfind('.')
start_epoch = int(cfg.TRAIN.NET_G[istart:iend])

start_epoch = int(cfg.TRAIN.NET_G[istart:iend]) lien giving the error which is calling bird.yml file

cfg/bird.yml file where "NET_G" is defined
TRAIN:
NF: 64 # default 64
BATCH_SIZE: 24 #24
MAX_EPOCH: 600
NET_G: '' # when validation, put the path of the trained model here
WARMUP_EPOCHS: 100
GSAVE_INTERVAL: 10
DSAVE_INTERVAL: 10

which trained model path should go in NET_G ? we have 2 trained model folders: fixed.zip and finetune.zip

should it be : fixed trained model

NET_G: 'trained_model/fixed/cub/netG_590.pth'

or finetune trained model

NET_G: 'trained_model/finetune/cub/which_file'
where which_file has 4 models: netG_550.pth, netD_550.pth, image_encoder_550.pth, text_encoder_550.pth,

if we are chosing finetune model then which model from the above 4 I should choose?

Sorry for long message

[cookie-ke]

so when I run main.py. I get this

Total filenames: 11788 001.Black_footed_Albatross/Black_Footed_Albatross_0046_18.jpg Load filenames from: data/birds/train/filenames.pickle (8855) Load filenames from: data/birds/test/filenames.pickle (2933) Load from: data/birds/captions.pickle 5450 10 /opt/conda/lib/python3.6/site-packages/torch/nn/modules/rnn.py:61: UserWarning: dropout option adds dropout after all but last recurrent layer, so non-zero dropout expects num_layers greater than 1, but got dropout=0.5 and num_layers=1 "num_layers={}".format(dropout, num_layers)) Downloading: "https://download.pytorch.org/models/inception_v3_google-1a9a5a14.pth" to /root/.cache/torch/hub/checkpoints/inception_v3_google-1a9a5a14.pth 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 104M/104M [00:01<00:00, 64.4MB/s] Load pretrained inception v3 model Load image encoder from: DAMSMencoders/bird/image_encoder200.pth Traceback (most recent call last): File "main.py", line 547, in sampling(text_encoder, netG, dataloader, ixtoword, device) # generate images for the whole valid dataset File "main.py", line 76, in sampling start_epoch = int(cfg.TRAIN.NET_G[istart:iend]) ValueError: invalid literal for int() with base 10: ''

this is the main.py def sampling(text_encoder, netG, dataloader, ixtoword, device): model_dir = cfg.TRAIN.NET_G istart = cfg.TRAIN.NET_G.rfind('_') + 1 iend = cfg.TRAIN.NET_G.rfind('.') start_epoch = int(cfg.TRAIN.NET_G[istart:iend])

start_epoch = int(cfg.TRAIN.NET_G[istart:iend]) lien giving the error which is calling bird.yml file

cfg/bird.yml file where "NET_G" is defined TRAIN: NF: 64 # default 64 BATCH_SIZE: 24 #24 MAX_EPOCH: 600 NET_G: '' # when validation, put the path of the trained model here WARMUP_EPOCHS: 100 GSAVE_INTERVAL: 10 DSAVE_INTERVAL: 10

which trained model path should go in NET_G ? we have 2 trained model folders: fixed.zip and finetune.zip

should it be : fixed trained model

NET_G: 'trained_model/fixed/cub/netG_590.pth'

or finetune trained model

NET_G: 'trained_model/finetune/cub/which_file' where which_file has 4 models: netG_550.pth, netD_550.pth, image_encoder_550.pth, text_encoder_550.pth,

if we are chosing finetune model then which model from the above 4 I should choose?

Sorry for long message

Sorry, I've never had that problem.

[priyankaupadhyay090]

cfg/bird.yml file where "NET_G" is defined
TRAIN:
NF: 64 # default 64
BATCH_SIZE: 24 #24
MAX_EPOCH: 600
NET_G: '' # when validation, put the path of the trained model here
WARMUP_EPOCHS: 100
GSAVE_INTERVAL: 10
DSAVE_INTERVAL: 10

okay. can you let me know which path you passed into NET_G for bird.yml file?

NET_G: '' # when validation, put the path of the trained model here

[cookie-ke]

cfg/bird.yml file where "NET_G" is defined
TRAIN:
NF: 64 # default 64
BATCH_SIZE: 24 #24
MAX_EPOCH: 600
NET_G: '' # when validation, put the path of the trained model here
WARMUP_EPOCHS: 100
GSAVE_INTERVAL: 10
DSAVE_INTERVAL: 10

okay. can you let me know which path you passed into NET_G for bird.yml file?

NET_G: '' # when validation, put the path of the trained model here

I'm not using this data set, so it may be structurally inconsistent.

[priyankaupadhyay090]

cfg/bird.yml file where "NET_G" is defined
TRAIN:
NF: 64 # default 64
BATCH_SIZE: 24 #24
MAX_EPOCH: 600
NET_G: '' # when validation, put the path of the trained model here
WARMUP_EPOCHS: 100
GSAVE_INTERVAL: 10
DSAVE_INTERVAL: 10

okay. can you let me know which path you passed into NET_G for bird.yml file?
NET_G: '' # when validation, put the path of the trained model here

I'm not using this data set, so it may be structurally inconsistent.

thank you :)