the train error in my dataset, the dimensions in the model are torch.Size([1]) and whose dimensions in the checkpoint are torch.Size([80]).
Opened this issue · 1 comments
Dear all:
when I train my data, the category count is 1,my code is modified as follow:
1.modified one:
models/CornerNet.py: line 72
class model(kp):
def init(self, db):
n = 5
dims = [256, 256, 384, 384, 384, 512]
modules = [2, 2, 2, 2, 2, 4]
out_dim = 1<---------------------- here need change to my category count
2.modified two:
db/detection.py: line 8
self._configs["categories"] = 1 <---------------------- here need change to my category count
3. modified three:
config/CornerNet.json: line 45
"categories": 1 <---------------------- here need change to my category count
4 the error is reported as floow,:
loading all datasets...
using 4 threads
loading from cache file: ./cache/coco_trainval2017.pkl
loading annotations into memory...
Done (t=0.13s)
creating index...
index created!
loading from cache file: ./cache/coco_trainval2017.pkl
loading annotations into memory...
Done (t=0.15s)
creating index...
index created!
loading from cache file: ./cache/coco_trainval2017.pkl
loading annotations into memory...
Done (t=0.17s)
creating index...
index created!
loading from cache file: ./cache/coco_trainval2017.pkl
loading annotations into memory...
Done (t=0.18s)
creating index...
index created!
loading from cache file: ./cache/coco_test2017.pkl
loading annotations into memory...
Done (t=0.12s)
creating index...
index created!
system config...
{'batch_size': 4,
'cache_dir': './cache',
'chunk_sizes': [4],
'config_dir': './config',
'data_dir': './data',
'data_rng': <mtrand.RandomState object at 0x7f28e24bd708>,
'dataset': 'MSCOCO',
'decay_rate': 10,
'display': 5,
'learning_rate': 0.00025,
'max_iter': 500000,
'nnet_rng': <mtrand.RandomState object at 0x7f28e24bd750>,
'opt_algo': 'adam',
'prefetch_size': 5,
'pretrain': './cache/nnet/CornerNet/CornerNet_500000.pkl',
'result_dir': './results',
'sampling_function': 'kp_detection',
'snapshot': 5000,
'snapshot_name': 'CornerNet',
'stepsize': 450000,
'test_split': 'testdev',
'train_split': 'trainval',
'val_iter': 100,
'val_split': 'minival',
'weight_decay': False,
'weight_decay_rate': 1e-05,
'weight_decay_type': 'l2'}
db config...
{'ae_threshold': 0.5,
'border': 128,
'categories': 1,
'data_aug': True,
'gaussian_bump': True,
'gaussian_iou': 0.3,
'gaussian_radius': -1,
'input_size': [511, 511],
'lighting': True,
'max_per_image': 100,
'merge_bbox': False,
'nms_algorithm': 'exp_soft_nms',
'nms_kernel': 3,
'nms_threshold': 0.5,
'output_sizes': [[128, 128]],
'rand_color': True,
'rand_crop': True,
'rand_pushes': False,
'rand_samples': False,
'rand_scale_max': 1.4,
'rand_scale_min': 0.6,
'rand_scale_step': 0.1,
'rand_scales': array([0.6, 0.7, 0.8, 0.9, 1. , 1.1, 1.2, 1.3]),
'special_crop': False,
'test_scales': [1],
'top_k': 100,
'weight_exp': 8}
len of db: 10615
start prefetching data...
start prefetching data...
start prefetching data...
shuffling indices...
start prefetching data...
start prefetching data...
shuffling indices...
shuffling indices...
shuffling indices...
building model...
module_file: models.CornerNet
shuffling indices...
total parameters: 200954000
loading from pretrained model
loading from ./cache/nnet/CornerNet/CornerNet_500000.pkl
Traceback (most recent call last):
File "/home/hhd/deeplearning/detection/github/CornerNet/train.py", line 195, in
train(training_dbs, validation_db, args.start_iter)
File "/home/hhd/deeplearning/detection/github/CornerNet/train.py", line 120, in train
nnet.load_pretrained_params(pretrained_model)
File "/home/hhd/deeplearning/detection/github/CornerNet/nnet/py_factory.py", line 110, in load_pretrained_params
self.model.load_state_dict(params)
File "/home/gxm/anaconda3/envs/CornerNet/lib/python3.6/site-packages/torch/nn/modules/module.py", line 721, in load_state_dict
self.class.name, "\n\t".join(error_msgs)))
RuntimeError: Error(s) in loading state_dict for DummyModule:
While copying the parameter named "module.tl_heats.0.1.weight", whose dimensions in the model are torch.Size([1, 256, 1, 1]) and whose dimensions in the checkpoint are torch.Size([80, 256, 1, 1]).
While copying the parameter named "module.tl_heats.0.1.bias", whose dimensions in the model are torch.Size([1]) and whose dimensions in the checkpoint are torch.Size([80]).
While copying the parameter named "module.tl_heats.1.1.weight", whose dimensions in the model are torch.Size([1, 256, 1, 1]) and whose dimensions in the checkpoint are torch.Size([80, 256, 1, 1]).
While copying the parameter named "module.tl_heats.1.1.bias", whose dimensions in the model are torch.Size([1]) and whose dimensions in the checkpoint are torch.Size([80]).
While copying the parameter named "module.br_heats.0.1.weight", whose dimensions in the model are torch.Size([1, 256, 1, 1]) and whose dimensions in the checkpoint are torch.Size([80, 256, 1, 1]).
While copying the parameter named "module.br_heats.0.1.bias", whose dimensions in the model are torch.Size([1]) and whose dimensions in the checkpoint are torch.Size([80]).
While copying the parameter named "module.br_heats.1.1.weight", whose dimensions in the model are torch.Size([1, 256, 1, 1]) and whose dimensions in the checkpoint are torch.Size([80, 256, 1, 1]).
While copying the parameter named "module.br_heats.1.1.bias", whose dimensions in the model are torch.Size([1]) and whose dimensions in the checkpoint are torch.Size([80]).
In py_factory.py, "def load_pretrained_params(self, pretrained_model)" should be modified like this:
def load_pretrained_params(self, pretrained_model):
print("loading from {}".format(pretrained_model))
with open(pretrained_model, "rb") as f:
params = torch.load(f)
params = {k: v for k, v in params.items()
if k in self.model.state_dict()
and self.model.state_dict()[k].shape == v.shape}
self.model.load_state_dict(params, strict=False)