error instantiating model

[maidmentdaniel]

Hi all,

I'm attempting to train mask-rcnn on a custom dataset. I'm new to the detectron2 framework, so please bare with me.

I'm registering a dataset and calling a baseline config as follows:

register_coco_instances("seg_train", {}, train_annot, train_src)
register_coco_instances("seg_validate", {}, validate_annot, validate_src)
cfg = get_config('new_baselines/mask_rcnn_R_50_FPN_100ep_LSJ.py', trained=True)
cfg.dataloader.train.dataset.names = "seg_train"
cfg.dataloader.test.dataset.names = "seg_validate"
default_setup(cfg, args)

Then running do_train as taken from tools/lazyconfig_train_net.py

model = instantiate(cfg.model)

As a result I get the following Traceback:

Traceback (most recent call last):
  File "train_net.py", line 117, in <module>
    launch(
  File "/home/dmaidment/miniconda3/envs/tron/lib/python3.8/site-packages/detectron2/engine/launch.py", line 82, in launch
    main_func(*args)
  File "train_net.py", line 106, in main
    do_train(args, cfg)
  File "train_net.py", line 53, in do_train
    model = instantiate(cfg.model)
  File "/home/dmaidment/miniconda3/envs/tron/lib/python3.8/site-packages/detectron2/config/instantiate.py", line 61, in instantiate
    cfg = {k: instantiate(v) for k, v in cfg.items()}
  File "/home/dmaidment/miniconda3/envs/tron/lib/python3.8/site-packages/detectron2/config/instantiate.py", line 61, in <dictcomp>
    cfg = {k: instantiate(v) for k, v in cfg.items()}
  File "/home/dmaidment/miniconda3/envs/tron/lib/python3.8/site-packages/detectron2/config/instantiate.py", line 61, in instantiate
    cfg = {k: instantiate(v) for k, v in cfg.items()}
  File "/home/dmaidment/miniconda3/envs/tron/lib/python3.8/site-packages/detectron2/config/instantiate.py", line 61, in <dictcomp>
    cfg = {k: instantiate(v) for k, v in cfg.items()}
  File "/home/dmaidment/miniconda3/envs/tron/lib/python3.8/site-packages/detectron2/config/instantiate.py", line 77, in instantiate
    return cls(**cfg)
  File "/home/dmaidment/miniconda3/envs/tron/lib/python3.8/site-packages/detectron2/config/config.py", line 192, in wrapped
    init_func(self, *args, **kwargs)
  File "/home/dmaidment/miniconda3/envs/tron/lib/python3.8/site-packages/detectron2/modeling/roi_heads/box_head.py", line 49, in __init__
    self._output_size = (input_shape.channels, input_shape.height, input_shape.width)
  File "/home/dmaidment/miniconda3/envs/tron/lib/python3.8/site-packages/omegaconf/listconfig.py", line 181, in __getattr__
    self._format_and_raise(
  File "/home/dmaidment/miniconda3/envs/tron/lib/python3.8/site-packages/omegaconf/base.py", line 231, in _format_and_raise
    format_and_raise(
  File "/home/dmaidment/miniconda3/envs/tron/lib/python3.8/site-packages/omegaconf/_utils.py", line 873, in format_and_raise
    _raise(ex, cause)
  File "/home/dmaidment/miniconda3/envs/tron/lib/python3.8/site-packages/omegaconf/_utils.py", line 771, in _raise
    raise ex.with_traceback(sys.exc_info()[2])  # set env var OC_CAUSE=1 for full trace
omegaconf.errors.ConfigAttributeError: ListConfig does not support attribute access
    full_key: [channels]
    object_type=list

My environment is as follows:

sys.platform           linux
Python                 3.8.0 (default, Nov  6 2019, 21:49:08) [GCC 7.3.0]
numpy                  1.22.3
detectron2             0.6
Compiler               GCC 7.3
CUDA compiler          not available
DETECTRON2_ENV_MODULE  <not set>
PyTorch                1.8.2
PyTorch debug build    False
GPU available          No: torch.cuda.is_available() == False
Pillow                 9.0.1
torchvision            0.9.2
fvcore                 0.1.5.post20220512
iopath                 0.1.9
cv2                    Not found

I'm aware that the environment is pretty sparse, this is just for basic dev, not that actual training.