evaluate cambridge error

[luocha0107]

(gs_model) root@autodl-container-f34d45a126-5b9fd9ad:~/data_user/ysl/DeViLoc# python evaluate.py configs/cambridge.yml --ckpt_path pretrained/deviloc_weights.ckpt
ERROR:albumentations.check_version:Error fetching version info
Traceback (most recent call last):
File "/root/miniconda3/envs/gs_model/lib/python3.9/site-packages/albumentations/check_version.py", line 29, in fetch_version_info
with opener.open(url, timeout=2) as response:
File "/root/miniconda3/envs/gs_model/lib/python3.9/urllib/request.py", line 517, in open
response = self._open(req, data)
File "/root/miniconda3/envs/gs_model/lib/python3.9/urllib/request.py", line 534, in _open
result = self._call_chain(self.handle_open, protocol, protocol +
File "/root/miniconda3/envs/gs_model/lib/python3.9/urllib/request.py", line 494, in _call_chain
result = func(*args)
File "/root/miniconda3/envs/gs_model/lib/python3.9/urllib/request.py", line 1389, in https_open
return self.do_open(http.client.HTTPSConnection, req,
File "/root/miniconda3/envs/gs_model/lib/python3.9/urllib/request.py", line 1350, in do_open
r = h.getresponse()
File "/root/miniconda3/envs/gs_model/lib/python3.9/http/client.py", line 1377, in getresponse
response.begin()
File "/root/miniconda3/envs/gs_model/lib/python3.9/http/client.py", line 339, in begin
self.headers = self.msg = parse_headers(self.fp)
File "/root/miniconda3/envs/gs_model/lib/python3.9/http/client.py", line 236, in parse_headers
headers = _read_headers(fp)
File "/root/miniconda3/envs/gs_model/lib/python3.9/http/client.py", line 216, in _read_headers
line = fp.readline(_MAXLINE + 1)
File "/root/miniconda3/envs/gs_model/lib/python3.9/socket.py", line 704, in readinto
return self._sock.recv_into(b)
File "/root/miniconda3/envs/gs_model/lib/python3.9/ssl.py", line 1275, in recv_into
return self.read(nbytes, buffer)
File "/root/miniconda3/envs/gs_model/lib/python3.9/ssl.py", line 1133, in read
return self._sslobj.read(len, buffer)
socket.timeout: The read operation timed out
{'batch_size': 1,
'ckpt_path': 'pretrained/deviloc_weights.ckpt',
'covis_clustering': False,
'dataset': 'aachen',
'main_cfg_path': 'configs/cambridge.yml',
'num_workers': 4,
'out_dir': 'deviloc_outputs',
'out_file': 'aachen_v11_eval.txt'}
2024-08-17 16:31:26.794 | INFO | main::220 - Args and config initialized!
2024-08-17 16:31:26.794 | INFO | main::221 - Do covisibily clustering: False
Traceback (most recent call last):
File "/root/data_user/ysl/DeViLoc/evaluate.py", line 227, in
model = Dense2D3DMatcher(config=config["model"])
File "/root/data_user/ysl/DeViLoc/deviloc/models/model.py", line 26, in init
self.matcher.load_state_dict(pretrained_matcher["state_dict"])
File "/root/data_user/ysl/DeViLoc/third_party/feat_matcher/TopicFM/src/models/topic_fm.py", line 90, in load_state_dict
return super().load_state_dict(state_dict, *args, **kwargs)
File "/root/miniconda3/envs/gs_model/lib/python3.9/site-packages/torch/nn/modules/module.py", line 2041, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for TopicFM:
Missing key(s) in state_dict: "backbone.conv1.weight", "backbone.layer1.dwconv.weight", "backbone.layer1.dwconv.bias", "backbone.layer1.pwconv1.weight", "backbone.layer1.pwconv1.bias", "backbone.layer1.norm2.gamma", "backbone.layer1.norm2.beta", "backbone.layer1.pwconv2.weight", "backbone.layer1.pwconv2.bias", "backbone.layer2.0.0.weight", "backbone.layer2.1.dwconv.weight", "backbone.layer2.1.dwconv.bias", "backbone.layer2.1.pwconv1.weight", "backbone.layer2.1.pwconv1.bias", "backbone.layer2.1.norm2.gamma", "backbone.layer2.1.norm2.beta", "backbone.layer2.1.pwconv2.weight", "backbone.layer2.1.pwconv2.bias", "backbone.layer3.0.0.weight", "backbone.layer3.1.dwconv.weight", "backbone.layer3.1.dwconv.bias", "backbone.layer3.1.pwconv1.weight", "backbone.layer3.1.pwconv1.bias", "backbone.layer3.1.norm2.gamma", "backbone.layer3.1.norm2.beta", "backbone.layer3.1.pwconv2.weight", "backbone.layer3.1.pwconv2.bias", "backbone.layer4.0.0.weight", "backbone.layer4.1.dwconv.weight", "backbone.layer4.1.dwconv.bias", "backbone.layer4.1.pwconv1.weight", "backbone.layer4.1.pwconv1.bias", "backbone.layer4.1.norm2.gamma", "backbone.layer4.1.norm2.beta", "backbone.layer4.1.pwconv2.weight", "backbone.layer4.1.pwconv2.bias", "backbone.layer3_outconv2.0.dwconv.weight", "backbone.layer3_outconv2.0.dwconv.bias", "backbone.layer3_outconv2.0.pwconv1.weight", "backbone.layer3_outconv2.0.pwconv1.bias", "backbone.layer3_outconv2.0.norm2.gamma", "backbone.layer3_outconv2.0.norm2.beta", "backbone.layer3_outconv2.0.pwconv2.weight", "backbone.layer3_outconv2.0.pwconv2.bias", "backbone.layer3_outconv2.1.dwconv.weight", "backbone.layer3_outconv2.1.dwconv.bias", "backbone.layer3_outconv2.1.pwconv1.weight", "backbone.layer3_outconv2.1.pwconv1.bias", "backbone.layer3_outconv2.1.norm2.gamma", "backbone.layer3_outconv2.1.norm2.beta", "backbone.layer3_outconv2.1.pwconv2.weight", "backbone.layer3_outconv2.1.pwconv2.bias", "backbone.layer2_outconv2.0.dwconv.weight", "backbone.layer2_outconv2.0.dwconv.bias", "backbone.layer2_outconv2.0.pwconv1.weight", "backbone.layer2_outconv2.0.pwconv1.bias", "backbone.layer2_outconv2.0.norm2.gamma", "backbone.layer2_outconv2.0.norm2.beta", "backbone.layer2_outconv2.0.pwconv2.weight", "backbone.layer2_outconv2.0.pwconv2.bias", "backbone.layer2_outconv2.1.dwconv.weight", "backbone.layer2_outconv2.1.dwconv.bias", "backbone.layer2_outconv2.1.pwconv1.weight", "backbone.layer2_outconv2.1.pwconv1.bias", "backbone.layer2_outconv2.1.norm2.gamma", "backbone.layer2_outconv2.1.norm2.beta", "backbone.layer2_outconv2.1.pwconv2.weight", "backbone.layer2_outconv2.1.pwconv2.bias", "backbone.layer1_outconv2.0.dwconv.weight", "backbone.layer1_outconv2.0.dwconv.bias", "backbone.layer1_outconv2.0.pwconv1.weight", "backbone.layer1_outconv2.0.pwconv1.bias", "backbone.layer1_outconv2.0.norm2.gamma", "backbone.layer1_outconv2.0.norm2.beta", "backbone.layer1_outconv2.0.pwconv2.weight", "backbone.layer1_outconv2.0.pwconv2.bias", "backbone.layer1_outconv2.1.dwconv.weight", "backbone.layer1_outconv2.1.dwconv.bias", "backbone.layer1_outconv2.1.pwconv1.weight", "backbone.layer1_outconv2.1.pwconv1.bias", "backbone.layer1_outconv2.1.norm2.gamma", "backbone.layer1_outconv2.1.norm2.beta", "backbone.layer1_outconv2.1.pwconv2.weight", "backbone.layer1_outconv2.1.pwconv2.bias", "fine_net.encoder_layers.2.mlp1.0.weight", "fine_net.encoder_layers.2.mlp1.0.bias", "fine_net.encoder_layers.2.mlp1.2.weight", "fine_net.encoder_layers.2.mlp1.2.bias", "fine_net.encoder_layers.2.mlp2.0.weight", "fine_net.encoder_layers.2.mlp2.0.bias", "fine_net.encoder_layers.2.mlp2.2.weight", "fine_net.encoder_layers.2.mlp2.2.bias", "fine_net.encoder_layers.2.norm1.weight", "fine_net.encoder_layers.2.norm1.bias", "fine_net.encoder_layers.2.norm2.weight", "fine_net.encoder_layers.2.norm2.bias", "fine_net.encoder_layers.3.mlp1.0.weight", "fine_net.encoder_layers.3.mlp1.0.bias", "fine_net.encoder_layers.3.mlp1.2.weight", "fine_net.encoder_layers.3.mlp1.2.bias", "fine_net.encoder_layers.3.mlp2.0.weight", "fine_net.encoder_layers.3.mlp2.0.bias", "fine_net.encoder_layers.3.mlp2.2.weight", "fine_net.encoder_layers.3.mlp2.2.bias", "fine_net.encoder_layers.3.norm1.weight", "fine_net.encoder_layers.3.norm1.bias", "fine_net.encoder_layers.3.norm2.weight", "fine_net.encoder_layers.3.norm2.bias".
Unexpected key(s) in state_dict: "backbone.layer0.conv.weight", "backbone.layer0.norm.weight", "backbone.layer0.norm.bias", "backbone.layer0.norm.running_mean", "backbone.layer0.norm.running_var", "backbone.layer0.norm.num_batches_tracked", "backbone.layer1.0.conv.weight", "backbone.layer1.0.norm.weight", "backbone.layer1.0.norm.bias", "backbone.layer1.0.norm.running_mean", "backbone.layer1.0.norm.running_var", "backbone.layer1.0.norm.num_batches_tracked", "backbone.layer1.1.conv.weight", "backbone.layer1.1.norm.weight", "backbone.layer1.1.norm.bias", "backbone.layer1.1.norm.running_mean", "backbone.layer1.1.norm.running_var", "backbone.layer1.1.norm.num_batches_tracked", "backbone.layer2.0.conv.weight", "backbone.layer2.0.norm.weight", "backbone.layer2.0.norm.bias", "backbone.layer2.0.norm.running_mean", "backbone.layer2.0.norm.running_var", "backbone.layer2.0.norm.num_batches_tracked", "backbone.layer2.1.conv.weight", "backbone.layer2.1.norm.weight", "backbone.layer2.1.norm.bias", "backbone.layer2.1.norm.running_mean", "backbone.layer2.1.norm.running_var", "backbone.layer2.1.norm.num_batches_tracked", "backbone.layer3.0.conv.weight", "backbone.layer3.0.norm.weight", "backbone.layer3.0.norm.bias", "backbone.layer3.0.norm.running_mean", "backbone.layer3.0.norm.running_var", "backbone.layer3.0.norm.num_batches_tracked", "backbone.layer3.1.conv.weight", "backbone.layer3.1.norm.weight", "backbone.layer3.1.norm.bias", "backbone.layer3.1.norm.running_mean", "backbone.layer3.1.norm.running_var", "backbone.layer3.1.norm.num_batches_tracked", "backbone.layer4.0.conv.weight", "backbone.layer4.0.norm.weight", "backbone.layer4.0.norm.bias", "backbone.layer4.0.norm.running_mean", "backbone.layer4.0.norm.running_var", "backbone.layer4.0.norm.num_batches_tracked", "backbone.layer4.1.conv.weight", "backbone.layer4.1.norm.weight", "backbone.layer4.1.norm.bias", "backbone.layer4.1.norm.running_mean", "backbone.layer4.1.norm.running_var", "backbone.layer4.1.norm.num_batches_tracked", "backbone.layer3_outconv2.0.conv.weight", "backbone.layer3_outconv2.0.norm.weight", "backbone.layer3_outconv2.0.norm.bias", "backbone.layer3_outconv2.0.norm.running_mean", "backbone.layer3_outconv2.0.norm.running_var", "backbone.layer3_outconv2.0.norm.num_batches_tracked", "backbone.layer3_outconv2.1.weight", "backbone.layer2_outconv2.0.conv.weight", "backbone.layer2_outconv2.0.norm.weight", "backbone.layer2_outconv2.0.norm.bias", "backbone.layer2_outconv2.0.norm.running_mean", "backbone.layer2_outconv2.0.norm.running_var", "backbone.layer2_outconv2.0.norm.num_batches_tracked", "backbone.layer2_outconv2.1.weight", "backbone.layer1_outconv2.0.conv.weight", "backbone.layer1_outconv2.0.norm.weight", "backbone.layer1_outconv2.0.norm.bias", "backbone.layer1_outconv2.0.norm.running_mean", "backbone.layer1_outconv2.0.norm.running_var", "backbone.layer1_outconv2.0.norm.num_batches_tracked", "backbone.layer1_outconv2.1.weight".
size mismatch for backbone.layer3_outconv.weight: copying a param with shape torch.Size([384, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([256, 384, 1, 1]).
size mismatch for backbone.layer2_outconv.weight: copying a param with shape torch.Size([256, 192, 1, 1]) from checkpoint, the shape in current model is torch.Size([128, 256, 1, 1]).
size mismatch for backbone.layer1_outconv.weight: copying a param with shape torch.Size([192, 128, 1, 1]) from checkpoint, the shape in current model is torch.Size([96, 128, 1, 1]).
size mismatch for backbone.norm_outlayer1.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for backbone.norm_outlayer1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.encoder_layers.0.mlp2.0.weight: copying a param with shape torch.Size([128, 128]) from checkpoint, the shape in current model is torch.Size([96, 96]).
size mismatch for fine_net.encoder_layers.0.mlp2.0.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.encoder_layers.0.mlp2.2.weight: copying a param with shape torch.Size([128, 128]) from checkpoint, the shape in current model is torch.Size([96, 96]).
size mismatch for fine_net.encoder_layers.0.mlp2.2.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.encoder_layers.0.norm1.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.encoder_layers.0.norm1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.encoder_layers.0.norm2.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.encoder_layers.0.norm2.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.encoder_layers.1.mlp2.0.weight: copying a param with shape torch.Size([128, 128]) from checkpoint, the shape in current model is torch.Size([96, 96]).
size mismatch for fine_net.encoder_layers.1.mlp2.0.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.encoder_layers.1.mlp2.2.weight: copying a param with shape torch.Size([128, 128]) from checkpoint, the shape in current model is torch.Size([96, 96]).
size mismatch for fine_net.encoder_layers.1.mlp2.2.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.encoder_layers.1.norm1.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.encoder_layers.1.norm1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.encoder_layers.1.norm2.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.encoder_layers.1.norm2.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.detector.0.mlp2.0.weight: copying a param with shape torch.Size([128, 128]) from checkpoint, the shape in current model is torch.Size([96, 96]).
size mismatch for fine_net.detector.0.mlp2.0.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.detector.0.mlp2.2.weight: copying a param with shape torch.Size([128, 128]) from checkpoint, the shape in current model is torch.Size([96, 96]).
size mismatch for fine_net.detector.0.mlp2.2.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.detector.0.norm1.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.detector.0.norm1.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.detector.0.norm2.weight: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.detector.0.norm2.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([96]).
size mismatch for fine_net.detector.1.weight: copying a param with shape torch.Size([1, 128]) from checkpoint, the shape in current model is torch.Size([1, 96]).

i test several times,but the problem seems to arise in the weight, network size mismatch

[TruongKhang]

Hi @luocha0107 , I updated the code of TopicFM in the dev_2 branch. So the pre-trained weights do not match with the TopicFM architecture. Let's switch TopicFM from the dev_2 to the main branch.

cd third_party/feat_matcher/TopicFM && git checkout main

[luocha0107]

thank you for reply.I have solved this problem.
Now,I want to do the test and inference part, that is, input the image and give the image pose, how do I do that.