TUI-NICR / ESANet

ESANet: Efficient RGB-D Semantic Segmentation for Indoor Scene Analysis

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

The following arguments are required:--ckpt_path

thaboub opened this issue · comments

commented

Hi Dear,
Great work!
I installed all required dependencies and datasets but when I am running through the below error/issue:
"eval.py: error: the following arguments are required: --ckpt_path". Please see details after running "python eval.py --dataset..."

I got this type of errors for all syntaxes that include ckpt_path. Any hint or recommendation to resolve this issue will be highly appreciated.
Thank you very much


(py39) PS C:\ESANet> python eval.py --dataset nyuv2 --dataset_dir ./src/datasets/nyuv2 --ckpt_path ./trained_models/nyuv2/r34_NBt1D.pth
2021-06-24 12:45:50.804733: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'cudart64_110.dll'; dlerror: cudart64_110.dll not found
usage: eval.py [-h] [--results_dir RESULTS_DIR] [--last_ckpt PATH] [--pretrained_dir PRETRAINED_DIR] [--pretrained_scenenet PRETRAINED_SCENENET]
[--no_imagenet_pretraining] [--finetune FINETUNE] [--freeze FREEZE] [--batch_size BATCH_SIZE]
[--batch_size_valid BATCH_SIZE_VALID] [--height HEIGHT] [--width WIDTH] [--epochs N] [--lr LR] [--weight_decay WEIGHT_DECAY]
[--momentum M] [--optimizer {SGD,Adam}] [--class_weighting {median_frequency,logarithmic,None}]
[--c_for_logarithmic_weighting C_FOR_LOGARITHMIC_WEIGHTING] [--he_init] [--valid_full_res] [--activation {relu,swish,hswish}]
[--encoder {resnet18,resnet34,resnet50}] [--encoder_block {BasicBlock,NonBottleneck1D}]
[--nr_decoder_blocks NR_DECODER_BLOCKS [NR_DECODER_BLOCKS ...]] [--encoder_depth {resnet18,resnet34,resnet50,None}]
[--modality {rgbd,rgb,depth}] [--encoder_decoder_fusion {add,None}] [--context_module {ppm,None,ppm-1-2-4-8,appm,appm-1-2-4-8}]
[--channels_decoder CHANNELS_DECODER] [--decoder_channels_mode {constant,decreasing}]
[--fuse_depth_in_rgb_encoder {SE-add,add,None}] [--upsampling {nearest,bilinear,learned-3x3,learned-3x3-zeropad}]
[--dataset {sunrgbd,nyuv2,cityscapes,cityscapes-with-depth,scenenetrgbd}] [--dataset_dir DATASET_DIR] [--raw_depth]
[--aug_scale_min AUG_SCALE_MIN] [--aug_scale_max AUG_SCALE_MAX] [-j N] [--debug] --ckpt_path CKPT_PATH
eval.py: error: the following arguments are required: --ckpt_path

Sorry, I can not reproduce your error. The error seems like you did not provide the argument --ckpt_path. But your command looks like it is there. You could try to remove required=True for this argument and check if it is parsed anyway.