Add code spell hook to pre-commit and fix typos (#1384)

* Add code-spell hook to pre-commit

* Add code-spell hook to pre-commit and fix typos

* fix setup.cfg

.pre-commit-config.yaml

@@ -35,6 +35,10 @@ repos:
       - id: markdownlint
         args: ["-r", "~MD002,~MD013,~MD029,~MD033,~MD034",
               "-t", "allow_different_nesting"]
+  - repo: https://github.com/codespell-project/codespell
+    rev: v2.1.0
+    hooks:
+      - id: codespell
   - repo: https://github.com/myint/docformatter
     rev: v1.3.1
     hooks:

README.md

@@ -43,7 +43,7 @@ There are two versions of MMCV:
 - **mmcv-full**: comprehensive, with full features and various CUDA ops out of box. It takes longer time to build.
 - **mmcv**: lite, without CUDA ops but all other features, similar to mmcv<1.0.0. It is useful when you do not need those CUDA ops.
 
-**Note**: Do not install both versions in the same environment, otherwise you may encounter errors like `ModuleNotFound`. You need to uninstall one before installing the other. `Installing the full verion is highly recommended if CUDA is avaliable`.
+**Note**: Do not install both versions in the same environment, otherwise you may encounter errors like `ModuleNotFound`. You need to uninstall one before installing the other. `Installing the full version is highly recommended if CUDA is available`.
 
 a. Install the full version.
 

docs/deployment/tensorrt_plugin.md

@@ -73,7 +73,7 @@ pip install $TENSORRT_DIR/onnx_graphsurgeon/onnx_graphsurgeon-0.2.6-py2.py3-none
 pip install $TENSORRT_DIR/graphsurgeon/graphsurgeon-0.4.5-py2.py3-none-any.whl
 ```
 
-For more detailed infomation of installing TensorRT using tar, please refer to [Nvidia' website](https://docs.nvidia.com/deeplearning/tensorrt/archives/tensorrt-721/install-guide/index.html#installing-tar).
+For more detailed information of installing TensorRT using tar, please refer to [Nvidia' website](https://docs.nvidia.com/deeplearning/tensorrt/archives/tensorrt-721/install-guide/index.html#installing-tar).
 
 #### Build on Linux
 

docs/get_started/build.md

@@ -189,7 +189,7 @@ pip list
    <... omit some lines ...>
    ```
 
-   This should already be done by CUDA installer. If not, or you have multiple version of CUDA tookit installed, set it with
+   This should already be done by CUDA installer. If not, or you have multiple version of CUDA toolkit installed, set it with
 
    ```shell
    $env:CUDA_HOME = "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.2"
@@ -202,7 +202,7 @@ pip list
    ```shell
    # Suppose you are using GTX 1080, which is of capability 6.1
    $env:TORCH_CUDA_ARCH_LIST="6.1"
-   # OR build all suppoted arch, will be slow
+   # OR build all supported arch, will be slow
    $env:TORCH_CUDA_ARCH_LIST="3.5 3.7 5.0 5.2 6.0 6.1 7.0 7.5"
    ```
 

docs/get_started/installation.md

@@ -6,7 +6,7 @@ There are two versions of MMCV:
 - **mmcv**: lite, without CUDA ops but all other features, similar to mmcv<1.0.0. It is useful when you do not need those CUDA ops.
 
 ```{warning}
-Do not install both versions in the same environment, otherwise you may encounter errors like `ModuleNotFound`. You need to uninstall one before installing the other. `Installing the full verion is highly recommended if CUDA is avaliable`.
+Do not install both versions in the same environment, otherwise you may encounter errors like `ModuleNotFound`. You need to uninstall one before installing the other. `Installing the full version is highly recommended if CUDA is avaliable`.
 ```
 
 a. Install the full version.

docs/understand_mmcv/cnn.md

@@ -140,7 +140,7 @@ We provide the following initialization methods.
 
 - kaiming_init
 
-  Initialize module paramters with the valuse according to the method
+  Initialize module parameters with the values according to the method
   described in [Delving deep into rectifiers: Surpassing human-level
   performance on ImageNet classification - He, K. et al. (2015)](https://www.cv-foundation.org/openaccess/content_iccv_2015/papers/He_Delving_Deep_into_ICCV_2015_paper.pdf)
 
@@ -430,7 +430,7 @@ Let us introduce the usage of `initialize` in detail.
     #           [2., 2., 2.],
     #           [2., 2., 2.]]]], requires_grad=True)
 
-    # inner init_cfg has highter priority
+    # inner init_cfg has higher priority
     model1 = FooConv1d(init_cfg1)
     model2 = FooConv2d(init_cfg2)
     init_cfg = dict(type='Constant', layer=['Conv1d', 'Conv2d'], val=4., bias=5.)
@@ -473,7 +473,7 @@ Let us introduce the usage of `initialize` in detail.
     #           [2., 2., 2.],
     #           [2., 2., 2.]]]], requires_grad=True)
 
-    # inner init_cfg has highter priority
+    # inner init_cfg has higher priority
     model1 = FooConv1d(init_cfg1)
     model2 = FooConv2d(init_cfg2)
     init_cfg = dict(type='Constant', layer=['Conv1d', 'Conv2d'], val=4., bias=5.)

docs/understand_mmcv/runner.md

@@ -92,7 +92,7 @@ We will walk you through the usage of runner with a classification task. The fol
 model=...
 # initialize optimizer, typically, we set: cfg.optimizer = dict(type='SGD', lr=0.1, momentum=0.9, weight_decay=0.0001)
 optimizer = build_optimizer(model, cfg.optimizer)
-# intialize the dataloader corresponding to the workflow(train/val)
+# initialize the dataloader corresponding to the workflow(train/val)
 data_loaders = [
         build_dataloader(
             ds,
@@ -119,7 +119,7 @@ runner = build_runner(
 **(3) Register training hooks and customized hooks.**
 
 ```python
-# register defalt hooks neccesary for traning
+# register default hooks necessary for training
 runner.register_training_hooks(
     # configs of learning rate, it is typically set as:
     # lr_config = dict(policy='step', step=[100, 150])

docs_zh_CN/understand_mmcv/cnn.md

@@ -419,7 +419,7 @@ conv = ConvModule(
     #           [2., 2., 2.],
     #           [2., 2., 2.]]]], requires_grad=True)
 
-    # inner init_cfg has highter priority
+    # inner init_cfg has higher priority
     model1 = FooConv1d(init_cfg1)
     model2 = FooConv2d(init_cfg2)
     init_cfg = dict(type='Constant', layer=['Conv1d', 'Conv2d'], val=4., bias=5.)
@@ -462,7 +462,7 @@ conv = ConvModule(
     #           [2., 2., 2.],
     #           [2., 2., 2.]]]], requires_grad=True)
 
-    # inner init_cfg has highter priority
+    # inner init_cfg has higher priority
     model1 = FooConv1d(init_cfg1)
     model2 = FooConv2d(init_cfg2)
     init_cfg = dict(type='Constant', layer=['Conv1d', 'Conv2d'], val=4., bias=5.)

mmcv/cnn/utils/weight_init.py

@@ -383,7 +383,7 @@ class UniformInit(BaseInit):
 
 @INITIALIZERS.register_module(name='Kaiming')
 class KaimingInit(BaseInit):
-    r"""Initialize module paramters with the valuse according to the method
+    r"""Initialize module parameters with the values according to the method
     described in `Delving deep into rectifiers: Surpassing human-level
     performance on ImageNet classification - He, K. et al. (2015).
     <https://www.cv-foundation.org/openaccess/content_iccv_2015/

mmcv/fileio/parse.py

@@ -4,7 +4,7 @@ def list_from_file(filename, prefix='', offset=0, max_num=0, encoding='utf-8'):
 
     Args:
         filename (str): Filename.
-        prefix (str): The prefix to be inserted to the begining of each item.
+        prefix (str): The prefix to be inserted to the beginning of each item.
         offset (int): The offset of lines.
         max_num (int): The maximum number of lines to be read,
             zeros and negatives mean no limitation.

mmcv/image/colorspace.py

@@ -86,7 +86,7 @@ def _convert_input_type_range(img):
 
     It converts the input image to np.float32 type and range of [0, 1].
     It is mainly used for pre-processing the input image in colorspace
-    convertion functions such as rgb2ycbcr and ycbcr2rgb.
+    conversion functions such as rgb2ycbcr and ycbcr2rgb.
 
     Args:
         img (ndarray): The input image. It accepts:
@@ -116,7 +116,7 @@ def _convert_output_type_range(img, dst_type):
     images will be converted to np.uint8 type with range [0, 255]. If
     `dst_type` is np.float32, it converts the image to np.float32 type with
     range [0, 1].
-    It is mainly used for post-processing images in colorspace convertion
+    It is mainly used for post-processing images in colorspace conversion
     functions such as rgb2ycbcr and ycbcr2rgb.
 
     Args:

mmcv/onnx/symbolic.py

@@ -51,7 +51,7 @@ def _interpolate(name, dim, interpolate_mode):
                 'Resize',
                 input,
                 empty_tensor,
-                # roi only takes effect whith
+                # roi only takes effect with
                 # coordinate_transformation_mode="tf_crop_and_resize"
                 scales,  # scales is not needed since we are sending out_size
                 output_size,

mmcv/ops/ball_query.py

@@ -22,7 +22,7 @@ class BallQuery(Function):
             center_xyz (Tensor): (B, npoint, 3) centers of the ball query.
 
         Returns:
-            Tensor: (B, npoint, nsample) tensor with the indicies of
+            Tensor: (B, npoint, nsample) tensor with the indices of
                 the features that form the query balls.
         """
         assert center_xyz.is_contiguous()

mmcv/ops/csrc/onnxruntime/cpu/gridSample.cpp

@@ -249,7 +249,7 @@ void GridSampleKernel::Compute(OrtKernelContext *context) {
           int64_t ix_nearest = static_cast<int64_t>(std::nearbyint(ix));
           int64_t iy_nearest = static_cast<int64_t>(std::nearbyint(iy));
 
-          // assign nearest neighor pixel value to output pixel
+          // assign nearest neighbor pixel value to output pixel
           float *out_ptr_NCHW = out_ptr + n * out_sN + h * out_sH + w * out_sW;
           const float *inp_ptr_NC = inp_ptr_N;
           for (int64_t c = 0; c < C;
@@ -284,7 +284,7 @@ void GridSampleKernel::Compute(OrtKernelContext *context) {
                ++c, out_ptr_NCHW += out_sC, inp_ptr_NC += inp_sC) {
             float coefficients[4];
 
-            // Interpolate 4 values in the x directon
+            // Interpolate 4 values in the x direction
             for (int64_t i = 0; i < 4; ++i) {
               coefficients[i] = cubic_interp1d<float>(
                   get_value_bounded<float>(inp_ptr_NC, ix_nw - 1, iy_nw - 1 + i,

mmcv/ops/csrc/onnxruntime/cpu/reduce_ops.cpp

@@ -122,7 +122,7 @@ void CumMax_CumMin_CPU(const T1 *input, T1 *output, T2 *indices,
           reversed_dim_cumprod[dim_i + 1] * out_dimensions.data()[dim_i];
     }
 
-    // do cummax or cummin besed on `Operation` type
+    // do cummax or cummin based on `Operation` type
     tensor_dim_apply3<float, int64_t>(
         input, output, indices, dim, ndims, reversed_dim_cumprod,
         cummax_cummin_helper<float, int64_t, Operation>);
@@ -135,7 +135,7 @@ void MMCVCumMaxKernel::Compute(OrtKernelContext *context) {
   const float *input_data =
       reinterpret_cast<const float *>(ort_.GetTensorData<float>(input));
 
-  // get ouput
+  // get output
   OrtTensorDimensions out_dimensions(ort_, input);
   OrtValue *output = ort_.KernelContext_GetOutput(
       context, 0, out_dimensions.data(), out_dimensions.size());
@@ -164,7 +164,7 @@ void MMCVCumMinKernel::Compute(OrtKernelContext *context) {
   const float *input_data =
       reinterpret_cast<const float *>(ort_.GetTensorData<float>(input));
 
-  // get ouput
+  // get output
   OrtTensorDimensions out_dimensions(ort_, input);
   OrtValue *output = ort_.KernelContext_GetOutput(
       context, 0, out_dimensions.data(), out_dimensions.size());

mmcv/ops/csrc/pytorch/info.cpp

@@ -25,7 +25,7 @@ std::string get_compiling_cuda_version() {
   printCudaStyleVersion(get_cudart_version());
   return oss.str();
 #else
-  return std::string("rocm not vailable");
+  return std::string("rocm not available");
 #endif
 #else
   return std::string("not available");

mmcv/ops/csrc/pytorch/modulated_deform_conv.cpp

@@ -99,10 +99,10 @@ void modulated_deform_conv_forward(
   const int kernel_w_ = weight.size(3);
 
   if (kernel_h_ != kernel_h || kernel_w_ != kernel_w)
-    AT_ERROR("Input shape and kernel shape wont match: (%d x %d vs %d x %d).",
+    AT_ERROR("Input shape and kernel shape won't match: (%d x %d vs %d x %d).",
              kernel_h_, kernel_w, kernel_h_, kernel_w_);
   if (channels != channels_kernel * group)
-    AT_ERROR("Input shape and kernel channels wont match: (%d vs %d).",
+    AT_ERROR("Input shape and kernel channels won't match: (%d vs %d).",
              channels, channels_kernel * group);
 
   const int height_out =
@@ -220,10 +220,10 @@ void modulated_deform_conv_backward(
   const int kernel_h_ = weight.size(2);
   const int kernel_w_ = weight.size(3);
   if (kernel_h_ != kernel_h || kernel_w_ != kernel_w)
-    AT_ERROR("Input shape and kernel shape wont match: (%d x %d vs %d x %d).",
+    AT_ERROR("Input shape and kernel shape won't match: (%d x %d vs %d x %d).",
              kernel_h_, kernel_w, kernel_h_, kernel_w_);
   if (channels != channels_kernel * group)
-    AT_ERROR("Input shape and kernel channels wont match: (%d vs %d).",
+    AT_ERROR("Input shape and kernel channels won't match: (%d vs %d).",
              channels, channels_kernel * group);
 
   const int height_out =

mmcv/ops/csrc/tensorrt/plugins/trt_grid_sampler_kernel.cu

@@ -148,7 +148,7 @@ __global__ void grid_sampler_2d_kernel(
     const int n = index / (out_H * out_W);
     const int grid_offset = n * grid_sN + h * grid_sH + w * grid_sW;
 
-    // get the corresponding input x, y co-ordinates from grid
+    // get the corresponding input x, y coordinates from grid
     scalar_t ix = grid[grid_offset];
     scalar_t iy = grid[grid_offset + grid_sCoor];
 
@@ -197,7 +197,7 @@ __global__ void grid_sampler_2d_kernel(
       int ix_nearest = static_cast<int>(::round(ix));
       int iy_nearest = static_cast<int>(::round(iy));
 
-      // assign nearest neighor pixel value to output pixel
+      // assign nearest neighbor pixel value to output pixel
       auto inp_ptr_NC = input + n * inp_sN;
       auto out_ptr_NCHW = output + n * out_sN + h * out_sH + w * out_sW;
       for (int c = 0; c < C;
@@ -249,7 +249,7 @@ __global__ void grid_sampler_3d_kernel(
     const int grid_offset =
         n * grid_sN + d * grid_sD + h * grid_sH + w * grid_sW;
 
-    // get the corresponding input x, y, z co-ordinates from grid
+    // get the corresponding input x, y, z coordinates from grid
     scalar_t ix = grid[grid_offset];
     scalar_t iy = grid[grid_offset + grid_sCoor];
     scalar_t iz = grid[grid_offset + 2 * grid_sCoor];
@@ -367,7 +367,7 @@ __global__ void grid_sampler_3d_kernel(
       int iy_nearest = static_cast<int>(::round(iy));
       int iz_nearest = static_cast<int>(::round(iz));
 
-      // assign nearest neighor pixel value to output pixel
+      // assign nearest neighbor pixel value to output pixel
       auto inp_ptr_NC = input + n * inp_sN;
       auto out_ptr_NCDHW =
           output + n * out_sN + d * out_sD + h * out_sH + w * out_sW;

mmcv/ops/csrc/tensorrt/plugins/trt_nms_kernel.cu

@@ -144,7 +144,7 @@ size_t get_onnxnms_workspace_size(size_t num_batches, size_t spatial_dimension,
  * @param[in] score_threshold threshold of scores
  * @param[in] offset box offset, only 0 or 1 is valid
  * @param[out] output with shape [output_length, 3], each row contain index
- *     (batch_id, class_id, boxes_id), filling -1 if result is not vaild.
+ *     (batch_id, class_id, boxes_id), filling -1 if result is not valid.
  * @param[in] center_point_box 0 if boxes is [left, top, right, bottom] 1 if
  *     boxes is [center_x, center_y, width, height]
  * @param[in] num_batches batch size of boxes and scores

mmcv/parallel/data_parallel.py

@@ -30,7 +30,7 @@ class MMDataParallel(DataParallel):
     def forward(self, *inputs, **kwargs):
         """Override the original forward function.
 
-        The main difference lies in the CPU inference where the datas in
+        The main difference lies in the CPU inference where the data in
         :class:`DataContainers` will still be gathered.
         """
         if not self.device_ids: