I keep getting error saying the list index out of range about flownet2-pytorch

[fitsumreda]

Remove the 'clean' from validation_dataset_root.
It will be added by mpisintelclean dataloader

[kimj09]

Thanks for your reply.
When I remove the 'clean' from validation_dataset_root,
I get this error saying,

main.py: error: argument insta--schedule_lr_frequency: invalid int value: 'FlyingChairs'

[fitsumreda]

To help debugging this, Can you post the complete command and stdouts here?

[kimj09]

This is my input :

python main.py --batch_size 8 --model FlowNet2C --optimizer=Adam --optimizer_lr=1e-4 --loss=MultiScale --loss_norm=L1
--loss_numScales=5 --loss_startScale=4 --optimizer_lr=1e-4 --crop_size 384 512
--training_dataset FlyingChairs --training_dataset_root /home/projects/flownet2-pytorch/FlyingChairs/FlyingChairs_release/data
--validation_dataset MpiSintelClean --validation_dataset_root /home/projects/flownet2-pytorch/MPI-Sintel-complete/training

the the output comes out as :

usage: main.py [-h] [--start_epoch START_EPOCH] [--total_epochs TOTAL_EPOCHS]
[--batch_size BATCH_SIZE] [--train_n_batches TRAIN_N_BATCHES]
[--crop_size CROP_SIZE [CROP_SIZE ...]]
[--gradient_clip GRADIENT_CLIP]
[--schedule_lr_fraction SCHEDULE_LR_FRACTION]
[--rgb_max RGB_MAX] [--number_workers NUMBER_WORKERS]
[--number_gpus NUMBER_GPUS] [--no_cuda] [--seed SEED]
[--name NAME] [--save SAVE]
[--validation_frequency VALIDATION_FREQUENCY]
[--validation_n_batches VALIDATION_N_BATCHES]
[--render_validation] [--inference]
[--inference_size INFERENCE_SIZE [INFERENCE_SIZE ...]]
[--inference_batch_size INFERENCE_BATCH_SIZE]
[--inference_n_batches INFERENCE_N_BATCHES] [--save_flow]
[--resume PATH] [--log_frequency LOG_FREQUENCY]
[--skip_training] [--skip_validation] [--fp16]
[--fp16_scale FP16_SCALE]
[--model {FlowNet2C,tofp32,FlowNet2CS,FlowNet2SD,ChannelNorm,FlowNet2CSS,tofp16,FlowNet2S,FlowNet2,Resample2d}]
insta--schedule_lr_frequency
main.py: error: argument insta--schedule_lr_frequency: invalid int value: 'FlyingChairs'

[fitsumreda]

i tested your command, it works fine for me.
I'm not able to reproduce your errors.
i'd suggest to pull again to make sure you've the latest updates.

[kimj09]

These are the steps I have made to run the code. Please tell me if I have missed any steps

get flownet2-pytorch source

git clone https://github.com/NVIDIA/flownet2-pytorch.git
cd flownet2-pytorch

install custom layers

bash install.sh

Build and launch docker image

bash launch_docker.sh

Example on MPISintel Final and Clean, with L1Loss on FlowNet2 model

python main.py --batch_size 8 --model FlowNet2 --loss=L1Loss --optimizer=Adam --optimizer_lr=1e-4
--training_dataset MpiSintelFinal --training_dataset_root home/projects/flownet2-pytorch/FlyingChairs/FlyingChairs_release/data
--validation_dataset MpiSintelClean --validation_dataset_root /home/projects/flownet2-pytorch/MPI-Sintel-complete/training

Are there any steps I am missing?

[fitsumreda]

That looks correct.
You may need to verify the data for flyingchair and mpisintel are indeed accessible inside the launched docker container.
I'd also make sure the python command is submitted as a single line command, they may be broken into several lines, and interpretted as several commands by bash or shell

[fitsumreda]

A correction, there is somthing off about your steps.
the bash install.sh must be done inside the container, i.e. after bash launch_docker.sh

[kimj09]

Thanks for your reply. I reinstalled and followed your direction as you pointed out but unfortunately, it keeps showing me
main.py: error: argument insta--schedule_lr_frequency: invalid int value: 'FlyingChairs'

I would have to figure it out what is wrong with mine.

[fitsumreda]

In python you can use, os.path.exists

[msieb1]

I am getting the same error.

"
File "/../libs/flownet2-pytorch/datasets.py", line 66, in __init__
self.frame_size = frame_utils.read_gen(self.image_list[0][0]).shape
IndexError: list index out of range
"

I am running

python main.py --inference --model FlowNet2 --save_flow --inference_dataset MpiSintelClean
--inference_dataset_root ../MPI-Sintel-testing/test/

[borkarra]

I am also getting this error. Does the MPI-Sintel dataset directory structure need to be modified in anyway after downloading? It seems like that would be the most likely source of this issue.

[TusharNimbhorkar]

@msieb1 I think you forgot to put the checkpoint/weights. (--resume /path/to/checkpoints)

@borkarra I don't think Sintel directory needs to be modified.

For inference
--inference_dataset_root should contain a folder name clean

Hope it works

[fitsumreda]

this is the folder structure of the opensource MPI-Sintel dataset that the code expects
e.g. --inference_dataset_root ./flow/training