[Feature] add onnxruntime test tool (#498)

* add onnxruntime test tool, update pytorch2onnx to support slice export

* onnx convert with custom output shape, update test code

* update pytorch2onnx, add rescale_shape support, add document

* update doc for lint error fixing

* remove cpu flag in ort_test.py

* change class name, fix cuda error

* remote comment

* fix bug of torch2onnx

* mIOU to mIoU

docs/useful_tools.md

@@ -53,6 +53,7 @@ python tools/pytorch2onnx.py \
     --output-file ${ONNX_FILE} \
     --input-img ${INPUT_IMG} \
     --shape ${INPUT_SHAPE} \
+    --rescale-shape ${RESCALE_SHAPE} \
     --show \
     --verify \
     --dynamic-export \
@@ -66,7 +67,8 @@ Description of arguments:
 - `--checkpoint` : The path of a model checkpoint file.
 - `--output-file`: The path of output ONNX model. If not specified, it will be set to `tmp.onnx`.
 - `--input-img` : The path of an input image for conversion and visualize.
-- `--shape`: The height and width of input tensor to the model. If not specified, it will be set to `256 256`.
+- `--shape`: The height and width of input tensor to the model. If not specified, it will be set to img_scale of testpipeline.
+- `--rescale-shape`: rescale shape of output, set this value to avoid OOM, only work on `slide` mode.
 - `--show`: Determines whether to print the architecture of the exported model. If not specified, it will be set to `False`.
 - `--verify`: Determines whether to verify the correctness of an exported model. If not specified, it will be set to `False`.
 - `--dynamic-export`: Determines whether to export ONNX model with dynamic input and output shapes. If not specified, it will be set to `False`.
@@ -74,6 +76,55 @@ Description of arguments:
 
 **Note**: This tool is still experimental. Some customized operators are not supported for now.
 
+### Evaluate ONNX model with ONNXRuntime
+
+We provide `tools/ort_test.py` to evaluate ONNX model with ONNXRuntime backend.
+
+#### Prerequisite
+
+- Install onnx and onnxruntime-gpu
+
+  ```shell
+  pip install onnx onnxruntime-gpu
+  ```
+
+#### Usage
+
+```python
+python tools/ort_test.py \
+    ${CONFIG_FILE} \
+    ${ONNX_FILE} \
+    --out ${OUTPUT_FILE} \
+    --eval ${EVALUATION_METRICS} \
+    --show \
+    --show-dir ${SHOW_DIRECTORY} \
+    --options ${CFG_OPTIONS} \
+    --eval-options ${EVALUATION_OPTIONS} \
+    --opacity ${OPACITY} \
+```
+
+Description of all arguments
+
+- `config`: The path of a model config file.
+- `model`: The path of a ONNX model file.
+- `--out`: The path of output result file in pickle format.
+- `--format-only` : Format the output results without perform evaluation. It is useful when you want to format the result to a specific format and submit it to the test server. If not specified, it will be set to `False`. Note that this argument is **mutually exclusive** with `--eval`.
+- `--eval`: Evaluation metrics, which depends on the dataset, e.g., "mIoU" for generic datasets, and "cityscapes" for Cityscapes. Note that this argument is **mutually exclusive** with `--format-only`.
+- `--show`: Show results flag.
+- `--show-dir`: Directory where painted images will be saved
+- `--options`: Override some settings in the used config file, the key-value pair in `xxx=yyy` format will be merged into config file.
+- `--eval-options`: Custom options for evaluation, the key-value pair in `xxx=yyy` format will be kwargs for `dataset.evaluate()` function
+- `--opacity`: Opacity of painted segmentation map. In (0, 1] range.
+
+#### Results and Models
+
+|   Model    |                     Config                     |  Dataset   | Metric | PyTorch | ONNXRuntime |
+| :--------: | :--------------------------------------------: | :--------: | :----: | :-----: | :---------: |
+|    FCN     |     fcn_r50-d8_512x1024_40k_cityscapes.py      | cityscapes |  mIoU  |  72.2   |    72.2     |
+|   PSPNet   |    pspnet_r50-d8_769x769_40k_cityscapes.py     | cityscapes |  mIoU  |  78.2   |    78.1     |
+| deeplabv3  |   deeplabv3_r50-d8_769x769_40k_cityscapes.py   | cityscapes |  mIoU  |  78.5   |    78.3     |
+| deeplabv3+ | deeplabv3plus_r50-d8_769x769_40k_cityscapes.py | cityscapes |  mIoU  |  78.9   |    78.7     |
+
 ### Convert to TorchScript (experimental)
 
 We also provide a script to convert model to [TorchScript](https://pytorch.org/docs/stable/jit.html) format. You can use the pytorch C++ API [LibTorch](https://pytorch.org/docs/stable/cpp_index.html) inference the trained model. The converted model could be visualized by tools like [Netron](https://github.com/lutzroeder/netron). Besides, we also support comparing the output results between Pytorch and TorchScript model.

tools/ort_test.py

@@ -0,0 +1,191 @@
+import argparse
+import os
+import os.path as osp
+import warnings
+
+import mmcv
+import numpy as np
+import onnxruntime as ort
+import torch
+from mmcv.parallel import MMDataParallel
+from mmcv.runner import get_dist_info
+from mmcv.utils import DictAction
+
+from mmseg.apis import single_gpu_test
+from mmseg.datasets import build_dataloader, build_dataset
+from mmseg.models.segmentors.base import BaseSegmentor
+
+
+class ONNXRuntimeSegmentor(BaseSegmentor):
+
+    def __init__(self, onnx_file, cfg, device_id):
+        super(ONNXRuntimeSegmentor, self).__init__()
+        # get the custom op path
+        ort_custom_op_path = ''
+        try:
+            from mmcv.ops import get_onnxruntime_op_path
+            ort_custom_op_path = get_onnxruntime_op_path()
+        except (ImportError, ModuleNotFoundError):
+            warnings.warn('If input model has custom op from mmcv, \
+                you may have to build mmcv with ONNXRuntime from source.')
+        session_options = ort.SessionOptions()
+        # register custom op for onnxruntime
+        if osp.exists(ort_custom_op_path):
+            session_options.register_custom_ops_library(ort_custom_op_path)
+        sess = ort.InferenceSession(onnx_file, session_options)
+        providers = ['CPUExecutionProvider']
+        options = [{}]
+        is_cuda_available = ort.get_device() == 'GPU'
+        if is_cuda_available:
+            providers.insert(0, 'CUDAExecutionProvider')
+            options.insert(0, {'device_id': device_id})
+
+        sess.set_providers(providers, options)
+
+        self.sess = sess
+        self.device_id = device_id
+        self.io_binding = sess.io_binding()
+        self.output_names = [_.name for _ in sess.get_outputs()]
+        for name in self.output_names:
+            self.io_binding.bind_output(name)
+        self.cfg = cfg
+        self.test_mode = cfg.model.test_cfg.mode
+
+    def extract_feat(self, imgs):
+        raise NotImplementedError('This method is not implemented.')
+
+    def encode_decode(self, img, img_metas):
+        raise NotImplementedError('This method is not implemented.')
+
+    def forward_train(self, imgs, img_metas, **kwargs):
+        raise NotImplementedError('This method is not implemented.')
+
+    def simple_test(self, img, img_meta, **kwargs):
+        device_type = img.device.type
+        self.io_binding.bind_input(
+            name='input',
+            device_type=device_type,
+            device_id=self.device_id,
+            element_type=np.float32,
+            shape=img.shape,
+            buffer_ptr=img.data_ptr())
+        self.sess.run_with_iobinding(self.io_binding)
+        seg_pred = self.io_binding.copy_outputs_to_cpu()[0]
+        # whole might support dynamic reshape
+        ori_shape = img_meta[0]['ori_shape']
+        if not (ori_shape[0] == seg_pred.shape[-2]
+                and ori_shape[1] == seg_pred.shape[-1]):
+            seg_pred = torch.from_numpy(seg_pred).float()
+            seg_pred = torch.nn.functional.interpolate(
+                seg_pred, size=tuple(ori_shape[:2]), mode='nearest')
+            seg_pred = seg_pred.long().detach().cpu().numpy()
+        seg_pred = seg_pred[0]
+        seg_pred = list(seg_pred)
+        return seg_pred
+
+    def aug_test(self, imgs, img_metas, **kwargs):
+        raise NotImplementedError('This method is not implemented.')
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(
+        description='mmseg onnxruntime backend test (and eval) a model')
+    parser.add_argument('config', help='test config file path')
+    parser.add_argument('model', help='Input model file')
+    parser.add_argument('--out', help='output result file in pickle format')
+    parser.add_argument(
+        '--format-only',
+        action='store_true',
+        help='Format the output results without perform evaluation. It is'
+        'useful when you want to format the result to a specific format and '
+        'submit it to the test server')
+    parser.add_argument(
+        '--eval',
+        type=str,
+        nargs='+',
+        help='evaluation metrics, which depends on the dataset, e.g., "mIoU"'
+        ' for generic datasets, and "cityscapes" for Cityscapes')
+    parser.add_argument('--show', action='store_true', help='show results')
+    parser.add_argument(
+        '--show-dir', help='directory where painted images will be saved')
+    parser.add_argument(
+        '--options', nargs='+', action=DictAction, help='custom options')
+    parser.add_argument(
+        '--eval-options',
+        nargs='+',
+        action=DictAction,
+        help='custom options for evaluation')
+    parser.add_argument(
+        '--opacity',
+        type=float,
+        default=0.5,
+        help='Opacity of painted segmentation map. In (0, 1] range.')
+    parser.add_argument('--local_rank', type=int, default=0)
+    args = parser.parse_args()
+    if 'LOCAL_RANK' not in os.environ:
+        os.environ['LOCAL_RANK'] = str(args.local_rank)
+    return args
+
+
+def main():
+    args = parse_args()
+
+    assert args.out or args.eval or args.format_only or args.show \
+        or args.show_dir, \
+        ('Please specify at least one operation (save/eval/format/show the '
+         'results / save the results) with the argument "--out", "--eval"'
+         ', "--format-only", "--show" or "--show-dir"')
+
+    if args.eval and args.format_only:
+        raise ValueError('--eval and --format_only cannot be both specified')
+
+    if args.out is not None and not args.out.endswith(('.pkl', '.pickle')):
+        raise ValueError('The output file must be a pkl file.')
+
+    cfg = mmcv.Config.fromfile(args.config)
+    if args.options is not None:
+        cfg.merge_from_dict(args.options)
+    cfg.model.pretrained = None
+    cfg.data.test.test_mode = True
+
+    # init distributed env first, since logger depends on the dist info.
+    distributed = False
+
+    # build the dataloader
+    # TODO: support multiple images per gpu (only minor changes are needed)
+    dataset = build_dataset(cfg.data.test)
+    data_loader = build_dataloader(
+        dataset,
+        samples_per_gpu=1,
+        workers_per_gpu=cfg.data.workers_per_gpu,
+        dist=distributed,
+        shuffle=False)
+
+    # load onnx config and meta
+    cfg.model.train_cfg = None
+    model = ONNXRuntimeSegmentor(args.model, cfg=cfg, device_id=0)
+    model.CLASSES = dataset.CLASSES
+    model.PALETTE = dataset.PALETTE
+
+    efficient_test = False
+    if args.eval_options is not None:
+        efficient_test = args.eval_options.get('efficient_test', False)
+
+    model = MMDataParallel(model, device_ids=[0])
+    outputs = single_gpu_test(model, data_loader, args.show, args.show_dir,
+                              efficient_test, args.opacity)
+
+    rank, _ = get_dist_info()
+    if rank == 0:
+        if args.out:
+            print(f'\nwriting results to {args.out}')
+            mmcv.dump(outputs, args.out)
+        kwargs = {} if args.eval_options is None else args.eval_options
+        if args.format_only:
+            dataset.format_results(outputs, **kwargs)
+        if args.eval:
+            dataset.evaluate(outputs, args.eval, **kwargs)
+
+
+if __name__ == '__main__':
+    main()

tools/pytorch2onnx.py

@@ -71,10 +71,13 @@ def _demo_mm_inputs(input_shape, num_classes):
     return mm_inputs
 
 
-def _prepare_input_img(img_path, test_pipeline, shape=None):
+def _prepare_input_img(img_path,
+                       test_pipeline,
+                       shape=None,
+                       rescale_shape=None):
     # build the data pipeline
     if shape is not None:
-        test_pipeline[1]['img_scale'] = shape
+        test_pipeline[1]['img_scale'] = (shape[1], shape[0])
     test_pipeline[1]['transforms'][0]['keep_ratio'] = False
     test_pipeline = [LoadImage()] + test_pipeline[1:]
     test_pipeline = Compose(test_pipeline)
@@ -84,6 +87,10 @@ def _prepare_input_img(img_path, test_pipeline, shape=None):
     imgs = data['img']
     img_metas = [i.data for i in data['img_metas']]
 
+    if rescale_shape is not None:
+        for img_meta in img_metas:
+            img_meta['ori_shape'] = tuple(rescale_shape) + (3, )
+
     mm_inputs = {'imgs': imgs, 'img_metas': img_metas}
 
     return mm_inputs
@@ -91,15 +98,24 @@ def _prepare_input_img(img_path, test_pipeline, shape=None):
 
 def _update_input_img(img_list, img_meta_list):
     # update img and its meta list
-    N, C, H, W = img_list[0].shape
+    N = img_list[0].size(0)
     img_meta = img_meta_list[0][0]
+    img_shape = img_meta['img_shape']
+    ori_shape = img_meta['ori_shape']
+    pad_shape = img_meta['pad_shape']
     new_img_meta_list = [[{
-        'img_shape': (H, W, C),
+        'img_shape':
-        'ori_shape': (H, W, C),
+        img_shape,
-        'pad_shape': (H, W, C),
+        'ori_shape':
-        'filename': img_meta['filename'],
+        ori_shape,
-        'scale_factor': 1.,
+        'pad_shape':
-        'flip': False,
+        pad_shape,
+        'filename':
+        img_meta['filename'],
+        'scale_factor':
+        (img_shape[1] / ori_shape[1], img_shape[0] / ori_shape[0]) * 2,
+        'flip':
+        False,
     } for _ in range(N)]]
 
     return img_list, new_img_meta_list
@@ -128,6 +144,7 @@ def pytorch2onnx(model,
             Default: False.
     """
     model.cpu().eval()
+    test_mode = model.test_cfg.mode
 
     if isinstance(model.decode_head, nn.ModuleList):
         num_classes = model.decode_head[-1].num_classes
@@ -136,18 +153,24 @@ def pytorch2onnx(model,
 
     imgs = mm_inputs.pop('imgs')
     img_metas = mm_inputs.pop('img_metas')
-    ori_shape = img_metas[0]['ori_shape']
 
     img_list = [img[None, :] for img in imgs]
     img_meta_list = [[img_meta] for img_meta in img_metas]
+    # update img_meta
     img_list, img_meta_list = _update_input_img(img_list, img_meta_list)
 
     # replace original forward function
     origin_forward = model.forward
     model.forward = partial(
-        model.forward, img_metas=img_meta_list, return_loss=False)
+        model.forward,
+        img_metas=img_meta_list,
+        return_loss=False,
+        rescale=True)
     dynamic_axes = None
     if dynamic_export:
+        if test_mode == 'slide':
+            dynamic_axes = {'input': {0: 'batch'}, 'output': {1: 'batch'}}
+        else:
             dynamic_axes = {
                 'input': {
                     0: 'batch',
@@ -182,7 +205,7 @@ def pytorch2onnx(model,
         onnx_model = onnx.load(output_file)
         onnx.checker.check_model(onnx_model)
 
-        if dynamic_export:
+        if dynamic_export and test_mode == 'whole':
             # scale image for dynamic shape test
             img_list = [
                 nn.functional.interpolate(_, scale_factor=1.5)
@@ -223,6 +246,10 @@ def pytorch2onnx(model,
             if not osp.exists(img):
                 img = imgs[0][:3, ...].permute(1, 2, 0) * 255
                 img = img.detach().numpy().astype(np.uint8)
+                ori_shape = img.shape[:2]
+            else:
+                ori_shape = LoadImage()({'img': img})['ori_shape']
+
             # resize onnx_result to ori_shape
             onnx_result_ = cv2.resize(onnx_result[0].astype(np.uint8),
                                       (ori_shape[1], ori_shape[0]))
@@ -271,8 +298,14 @@ def parse_args():
         '--shape',
         type=int,
         nargs='+',
-        default=[256, 256],
+        default=None,
-        help='input image size')
+        help='input image height and width.')
+    parser.add_argument(
+        '--rescale_shape',
+        type=int,
+        nargs='+',
+        default=None,
+        help='output image rescale height and width, work for slide mode.')
     parser.add_argument(
         '--cfg-options',
         nargs='+',
@@ -294,7 +327,15 @@ def parse_args():
 if __name__ == '__main__':
     args = parse_args()
 
-    if len(args.shape) == 1:
+    cfg = mmcv.Config.fromfile(args.config)
+    if args.cfg_options is not None:
+        cfg.merge_from_dict(args.cfg_options)
+    cfg.model.pretrained = None
+
+    if args.shape is None:
+        img_scale = cfg.test_pipeline[1]['img_scale']
+        input_shape = (1, 3, img_scale[1], img_scale[0])
+    elif len(args.shape) == 1:
         input_shape = (1, 3, args.shape[0], args.shape[0])
     elif len(args.shape) == 2:
         input_shape = (
@@ -304,10 +345,7 @@ if __name__ == '__main__':
     else:
         raise ValueError('invalid input shape')
 
-    cfg = mmcv.Config.fromfile(args.config)
+    test_mode = cfg.model.test_cfg.mode
-    if args.cfg_options is not None:
-        cfg.merge_from_dict(args.cfg_options)
-    cfg.model.pretrained = None
 
     # build the model and load checkpoint
     cfg.model.train_cfg = None
@@ -324,8 +362,15 @@ if __name__ == '__main__':
 
     # read input or create dummpy input
     if args.input_img is not None:
-        mm_inputs = _prepare_input_img(args.input_img, cfg.data.test.pipeline,
+        preprocess_shape = (input_shape[2], input_shape[3])
-                                       (input_shape[3], input_shape[2]))
+        rescale_shape = None
+        if args.rescale_shape is not None:
+            rescale_shape = [args.rescale_shape[0], args.rescale_shape[1]]
+        mm_inputs = _prepare_input_img(
+            args.input_img,
+            cfg.data.test.pipeline,
+            shape=preprocess_shape,
+            rescale_shape=rescale_shape)
     else:
         if isinstance(segmentor.decode_head, nn.ModuleList):
             num_classes = segmentor.decode_head[-1].num_classes