数据维度问题

[unlessyo]

class TrafficBase(Dataset):
def init(self,
data_root,
txt_file,
size = None,
):
self.data_root = data_root
self.data_paths = txt_file
with open(self.data_paths, "r") as f:
self.image_paths = f.read().splitlines()
self._length = len(self.image_paths)
self.size = size

def __len__(self):
    return self._length

def __getitem__(self, i):
    example = dict()
    path = self.image_paths[i]
    traffic_npy = np.load(path)

    # traffic_npy = np.array(traffic_npy).astype(np.uint8)
    traffic_npy[:,:,2][traffic_npy[:,:,2] > 3600] = 3600
    traffic_npy[:,:,1][traffic_npy[:,:,1] > 150] = 150

    traffic_npy[:,:,0] = (traffic_npy[:,:,0] / 5.0).astype(np.float32)
    traffic_npy[:,:,1] = (traffic_npy[:,:,1] / 150.0).astype(np.float32)
    traffic_npy[:,:,2] = (traffic_npy[:,:,2] / 3600.0).astype(np.float32)
    example['image'] = traffic_npy

    if 'train' in path:
        textpath = './datasets/traffic/train/' + 'text/' + path.split('/')[-1].split('.')[0] + '.txt'
    else:
        textpath = './datasets/traffic/validation/' + 'text/' + path.split('/')[-1].split('.')[0] + '.txt'
    with open(textpath, "r") as f:
        text = str(f.read().splitlines()[0])
    example['caption'] = text
    # example['structure'] = np.load('/home/zcy/latent-diffusion-main/datasets/traffic/matrix_roadclass&length.npy')
    return example

此函数中：
traffic_npy[:,:,0] = (traffic_npy[:,:,0] / 5.0).astype(np.float32)
traffic_npy[:,:,1] = (traffic_npy[:,:,1] / 150.0).astype(np.float32)
traffic_npy[:,:,2] = (traffic_npy[:,:,2] / 3600.0).astype(np.float32)
显示需要处理数据大小为[:,:,3]的数据。根据通信作者提供的数据集，.npy中的数据大小为[:,:,2]。请问第三维数据是什么？提供的代码或数据是否有误？

[ChyaZhang]

注释掉第三维即可

[yue0708]

return self.training_type_plugin.validation_step(*step_kwargs.values()) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/pytorch_lightning/plugins/training_type/ddp.py", line 386, in validation_step return self.model(*args, **kwargs) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl result = self.forward(*input, **kwargs) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/parallel/distributed.py", line 705, in forward output = self.module(*inputs[0], **kwargs[0]) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl result = self.forward(*input, **kwargs) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/pytorch_lightning/overrides/base.py", line 93, in forward output = self.module.validation_step(*inputs, **kwargs) File "/root/autodl-fs/ChatTraffic/ldm/models/autoencoder.py", line 374, in validation_step reconstructions, posterior = self(inputs) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl result = self.forward(*input, **kwargs) File "/root/autodl-fs/ChatTraffic/ldm/models/autoencoder.py", line 336, in forward posterior = self.encode(input) File "/root/autodl-fs/ChatTraffic/ldm/models/autoencoder.py", line 325, in encode h = self.encoder(x) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl result = self.forward(*input, **kwargs) File "/root/autodl-fs/ChatTraffic/ldm/modules/diffusionmodules/model.py", line 439, in forward hs = [self.conv_in(x)] File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl result = self.forward(*input, **kwargs) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/conv.py", line 399, in forward return self._conv_forward(input, self.weight, self.bias) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/conv.py", line 395, in _conv_forward return F.conv2d(input, weight, bias, self.stride, RuntimeError: Given groups=1, weight of size [128, 3, 3, 3], expected input[64, 2, 36, 36] to have 3 channels, but got 2 channels instead
我也有相似的问题，论文中提到的traffic data维度是36，36，3，3代表3个特征么，但是数据集的数据维度只有36，36，2，把代码的第三维注释了还是有报错，提示通道数不匹配，请问这是否是数据集不完整导致的呢

[ChyaZhang]

return self.training_type_plugin.validation_step(*step_kwargs.values()) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/pytorch_lightning/plugins/training_type/ddp.py", line 386, in validation_step return self.model(*args, **kwargs) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl result = self.forward(*input, **kwargs) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/parallel/distributed.py", line 705, in forward output = self.module(*inputs[0], **kwargs[0]) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl result = self.forward(*input, **kwargs) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/pytorch_lightning/overrides/base.py", line 93, in forward output = self.module.validation_step(*inputs, **kwargs) File "/root/autodl-fs/ChatTraffic/ldm/models/autoencoder.py", line 374, in validation_step reconstructions, posterior = self(inputs) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl result = self.forward(*input, **kwargs) File "/root/autodl-fs/ChatTraffic/ldm/models/autoencoder.py", line 336, in forward posterior = self.encode(input) File "/root/autodl-fs/ChatTraffic/ldm/models/autoencoder.py", line 325, in encode h = self.encoder(x) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl result = self.forward(*input, **kwargs) File "/root/autodl-fs/ChatTraffic/ldm/modules/diffusionmodules/model.py", line 439, in forward hs = [self.conv_in(x)] File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/module.py", line 889, in _call_impl result = self.forward(*input, **kwargs) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/conv.py", line 399, in forward return self._conv_forward(input, self.weight, self.bias) File "/root/miniconda3/envs/ChatTraffic/lib/python3.8/site-packages/torch/nn/modules/conv.py", line 395, in _conv_forward return F.conv2d(input, weight, bias, self.stride, RuntimeError: Given groups=1, weight of size [128, 3, 3, 3], expected input[64, 2, 36, 36] to have 3 channels, but got 2 channels instead 我也有相似的问题，论文中提到的traffic data维度是36，36，3，3代表3个特征么，但是数据集的数据维度只有36，36，2，把代码的第三维注释了还是有报错，提示通道数不匹配，请问这是否是数据集不完整导致的呢

发布的数据集确实没有第三个维度的数据，因为这个维度的数据经过我们检查在某些月份存在部分缺失，因此就没有提供。如果只是要预测速度数据的话，要解决这个问题最简单的方法就是可以把速度维度repeat3次。

[JsonL0]

论文中的输入数据x_i维度为(1260 * 2), 但在implementation detail小节将其重塑为36 * 36 * 2，reshape之后是1296 * 2, 和1260个顶点似乎对不上。我想请问数据维度这个36, 36, 2的数据维度分别代表的是什么？或者我可能漏掉了哪些信息？

[ChyaZhang]

论文中的输入数据x_i维度为(1260 * 2), 但在implementation detail小节将其重塑为36 * 36 * 2，reshape之后是1296 * 2, 和1260个顶点似乎对不上。我想请问数据维度这个36, 36, 2的数据维度分别代表的是什么？或者我可能漏掉了哪些信息？

1260是节点数，然后用0填充到1296再reshape，2个维度分别是速度数据和拥挤程度数据

[1013764208]

你好，是否可以提供下 有关包含第三个维度“流量”的 完整的数据集

[ChyaZhang]

你好，是否可以提供下 有关包含第三个维度“流量”的 完整的数据集

数据集中没有流量数据，只有速度和拥挤程度，第三个维度原本是通行时间，但是这个维度缺失比较严重就没放出来