Error while running image.py

[dhaval-zala-aivid]

cmd: python image.py DAD-COCOHead-ResNet50-best --is_full_range --config settings/DAD3DHeads_COCOHead.yml --network resnet50 --gpu_ids 0

ERROR:
train_labeled : DAD3DHeads
train_unlabeled : COCOHead
type_unsuper : ce
ulb_batch_ratio : 4.0
val_frequency : 1000
Traceback (most recent call last):
File "image.py", line 125, in
config = get_config('test')
File "/home/dhavalsinh/Desktop/research/face_direction/semiuhpe/src/config.py", line 195, in get_config
config = Config(phase)
File "/home/dhavalsinh/Desktop/research/face_direction/semiuhpe/src/config.py", line 37, in init
self.exp_detail, self.date, self.ckpt = self.test_ckpt.split('/')
ValueError: not enough values to unpack (expected 3, got 1)

I think there is some problem within the code

the same issue occurs with video.py

[hnuzhy]

You need to download the released weight file DAD-COCOHead-ResNet50-best from my huggingface link

[dhaval-zala-aivid]

@hnuzhy Again getting the same error...!!

stage1_iteration    : 50000
std_ratio           : 3.0
test_ckpt           : DAD-COCOHead-ResNet50-best.pth
test_set            : DAD3DHeads
train_labeled       : DAD3DHeads
train_unlabeled     : COCOHead
type_unsuper        : ce
ulb_batch_ratio     : 4.0
val_frequency       : 1000
Traceback (most recent call last):
  File "image.py", line 125, in <module>
    config = get_config('test')
  File "/home/dhavalsinh/Desktop/research/face_direction/semiuhpe/src/config.py", line 195, in get_config
    config = Config(phase)
  File "/home/dhavalsinh/Desktop/research/face_direction/semiuhpe/src/config.py", line 37, in __init__
    self.exp_detail, self.date, self.ckpt = self.test_ckpt.split('/')
ValueError: not enough values to unpack (expected 3, got 1)

[hnuzhy]

Maybe you need to change the test_ckpt into SSL1.0_r0.05_ce_effinetv2_tDyna0.75_b64_ema_RO_CO_CM_full/Julxx_xxxxxx/best, otherwise the self.exp_detail and self.date will not be parsed. Also remember to build the folder exps/SSL1.0_r0.05_ce_effinetv2_tDyna0.75_b64_ema_RO_CO_CM_full/Julxx_xxxxxx/ in your root path.

[hnuzhy]

For example, you can manually build a hierarchical folder with path `exps/DAD3DHeads_COCOHead/SSL1.0_r0.05_ce_effinetv2_tDyna0.75_b32_ema_RO_CO_CM_full/Jul18_100557/` under the root path `./SemiUHPE`. Then, you can put the downloaded file https://huggingface.co/HoyerChou/SemiUHPE/blob/main/DAD-WildHead-EffNetV2-S-best.pth into the new path. Finally, you may successfully run the inference script.