add onnx

research/cv/3dcnn/README_CN.md

@@ -139,6 +139,7 @@ BraTS 2017原始数据集的文件目录结构如下所示：
     │   ├── run_distribute_train_gpu.sh   # 启动GPU分布式训练（8卡）
     │   ├── run_standalone_train_gpu.sh   # 启动GPU单机训练（单卡）
     │   ├── run_eval_gpu.sh               # 启动GPU评估
+    │   ├── run_onnx_eval.sh              # ONNX推理shell脚本
     ├── src
     │   ├── config.py                     # yaml文件解析
     │   ├── dataset.py                    # 创建数据集
@@ -152,6 +153,7 @@ BraTS 2017原始数据集的文件目录结构如下所示：
     │   ├── train.txt                     # 训练数据集
     ├── train.py                          # 训练脚本
     ├── eval.py                           # 评估脚本
+    ├── eval_onnx.py                      # ONNX评估脚本
     ├── export.py                         # 推理模型导出脚本
     ├── preprocess.py                     # 310推理数据预处理
     ├── postprocess.py                    # 310推理数据后处理
@@ -303,7 +305,7 @@ mean dice enhance:
 python export.py --ckpt_file [CKPT_PATH] --file_name [FILE_NAME] --file_format [FILE_FORMAT]
 ```
 
-参数ckpt_file 是必需的，EXPORT_FORMAT 必须在 ["AIR", "MINDIR"]中进行选择。
+参数ckpt_file 是必需的，EXPORT_FORMAT 必须在 ["AIR", "MINDIR", "ONNX"]中进行选择。
 
 ## 推理过程
 
@@ -322,6 +324,21 @@ python export.py --ckpt_file [CKPT_PATH] --file_name [FILE_NAME] --file_format [
   bash run_infer_310.sh [MINDIR_PATH] [DATASET_PATH] [TEST_PATH] [DEVICE_ID]
   ```
 
+### ONNX推理
+
+在执行推理之前，需要通过export.py导出onnx文件
+
+- 在GPU上使用修正后的BraST 2017训练数据集进行推理
+
+  执行推理的命令如下所示，其中`DATA_PATH`是数据集路径；`TEST_PATH`是推理数据路径；`ONNX_PATH`是ONNX文件路径；`DEVICE_ID`可选，默认值为0。
+
+  ```shell
+  # ONNX 推理
+  bash run_onnx_eval.sh [DATA_PATH] [TEST_PATH] [ONNX_PATH] [CONFIG_PATH] [DEVICE_ID]
+  ```
+
+  上述python命令将在后台运行，您可以通过eval_onnx.log文件查看结果。
+
 # 模型描述
 
 ## 性能

research/cv/3dcnn/config.yaml

@@ -51,6 +51,9 @@ image_width: 240
 image_height: 240
 image_channel: 155
 
+# onnx eval
+onnx_path: ""
+
 ---
 # Help description for each configuration
 data_path: "The directory of data."

research/cv/3dcnn/config_gpu.yaml

@@ -40,6 +40,9 @@ ckpt_file: "./dense24-5_4200.ckpt"
 file_name: "dense24"
 file_format: "MINDIR"
 
+# onnx eval
+onnx_path: ""
+
 ---
 # Help description for each configuration
 data_path: "The directory of data."

research/cv/3dcnn/eval_onnx.py

+# Copyright 2022 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+"""
+python eval.py
+"""
+import numpy as np
+import onnxruntime as ort
+
+from src.config import config
+from src.dataset import vox_generator_test
+
+def one_hot(label, num_classes):
+    """ one-hot encode """
+    label_ = np.zeros([len(label), num_classes])
+    label_[np.arange(len(label)), label] = 1
+    return label_
+
+def calculate_dice(true_label, pred_label, num_classes):
+    """
+    calculate dice
+
+    Args:
+        true_label: true sparse labels
+        pred_label: predict sparse labels
+        num_classes: number of classes
+
+    Returns:
+        dice evaluation index
+    """
+    true_label = true_label.astype(int)
+    pred_label = pred_label.astype(int)
+    true_label = true_label.flatten()
+    true_label = one_hot(true_label, num_classes)
+    pred_label = pred_label.flatten()
+    pred_label = one_hot(pred_label, num_classes)
+    intersection = np.sum(true_label * pred_label, axis=0)
+    return (2. * intersection) / (np.sum(true_label, axis=0) + np.sum(pred_label, axis=0))
+
+def create_session(checkpoint_path, target_device):
+    '''
+    create onnx session
+    '''
+    if target_device == 'GPU':
+        providers = ['CUDAExecutionProvider']
+    elif target_device == 'CPU':
+        providers = ['CPUExecutionProvider']
+    else:
+        raise ValueError(f'Unsupported target device {target_device!r}. Expected one of: "CPU", "GPU"')
+    session = ort.InferenceSession(checkpoint_path, providers=providers)
+    input_name_1 = session.get_inputs()[0].name
+    input_name_2 = session.get_inputs()[1].name
+
+    return session, input_name_1, input_name_2
+
+if __name__ == '__main__':
+    # test dataset
+    test_files = []
+    with open(config.test_path) as f:
+        for line in f:
+            test_files.append(line[:-1])
+
+    data_gen_test = vox_generator_test(config.data_path, test_files, config.correction)
+
+    # network
+    network, flair_t2_node_name, t1_t1ce_node_name = create_session(config.onnx_path, config.device_target)
+
+    OFFSET_H = config.offset_height
+    OFFSET_W = config.offset_width
+    OFFSET_C = config.offset_channel
+    HSIZE = config.height_size
+    WSIZE = config.width_size
+    CSIZE = config.channel_size
+    PSIZE = config.pred_size
+
+    OFFSET_PH = (HSIZE - PSIZE) // 2
+    OFFSET_PW = (WSIZE - PSIZE) // 2
+    OFFSET_PC = (CSIZE - PSIZE) // 2
+
+    batches_w = int(np.ceil((240 - WSIZE) / float(OFFSET_W))) + 1
+    batches_h = int(np.ceil((240 - HSIZE) / float(OFFSET_H))) + 1
+    batches_c = int(np.ceil((155 - CSIZE) / float(OFFSET_C))) + 1
+
+    dice_whole, dice_core, dice_et = [], [], []
+    for i in range(len(test_files)):
+        print('predicting %s' % test_files[i])
+        x, x_n, y = data_gen_test.__next__()
+        pred = np.zeros([240, 240, 155, 5])
+        for hi in range(batches_h):
+            offset_h = min(OFFSET_H * hi, 240 - HSIZE)
+            offset_ph = offset_h + OFFSET_PH
+            for wi in range(batches_w):
+                offset_w = min(OFFSET_W * wi, 240 - WSIZE)
+                offset_pw = offset_w + OFFSET_PW
+                for ci in range(batches_c):
+                    offset_c = min(OFFSET_C * ci, 155 - CSIZE)
+                    offset_pc = offset_c + OFFSET_PC
+                    data = x[offset_h:offset_h + HSIZE, offset_w:offset_w + WSIZE, offset_c:offset_c + CSIZE, :]
+                    data_norm = x_n[offset_h:offset_h + HSIZE, offset_w:offset_w + WSIZE, offset_c:offset_c + CSIZE, :]
+                    data_norm = np.expand_dims(data_norm, 0)
+                    if not np.max(data) == 0 and np.min(data) == 0:
+                        flair_t2_node = data_norm[:, :, :, :, :2]
+                        t1_t1ce_node = data_norm[:, :, :, :, 2:]
+                        flair_t2_node = np.transpose(flair_t2_node, axes=[0, 4, 1, 2, 3])
+                        t1_t1ce_node = np.transpose(t1_t1ce_node, axes=[0, 4, 1, 2, 3])
+                        flair_t2_score, t1_t1ce_score = network.run(None, {flair_t2_node_name: flair_t2_node,
+                                                                           t1_t1ce_node_name: t1_t1ce_node})
+                        t1_t1ce_score = np.transpose(t1_t1ce_score, axes=[0, 2, 3, 4, 1])
+                        pred[offset_ph:offset_ph + PSIZE, offset_pw:offset_pw + PSIZE, offset_pc:offset_pc + PSIZE, :] \
+                            += np.squeeze(t1_t1ce_score)
+
+        pred = np.argmax(pred, axis=-1)
+        pred = pred.astype(int)
+        print('calculating dice...')
+        whole_pred = (pred > 0).astype(int)
+        whole_gt = (y > 0).astype(int)
+        core_pred = (pred == 1).astype(int) + (pred == 4).astype(int)
+        core_gt = (y == 1).astype(int) + (y == 4).astype(int)
+        et_pred = (pred == 4).astype(int)
+        et_gt = (y == 4).astype(int)
+        dice_whole_batch = calculate_dice(whole_gt, whole_pred, 2)
+        dice_core_batch = calculate_dice(core_gt, core_pred, 2)
+        dice_et_batch = calculate_dice(et_gt, et_pred, 2)
+        dice_whole.append(dice_whole_batch)
+        dice_core.append(dice_core_batch)
+        dice_et.append(dice_et_batch)
+        print(dice_whole_batch)
+        print(dice_core_batch)
+        print(dice_et_batch)
+
+    dice_whole = np.array(dice_whole)
+    dice_core = np.array(dice_core)
+    dice_et = np.array(dice_et)
+
+    print('mean dice whole:')
+    print(np.mean(dice_whole, axis=0))
+    print('mean dice core:')
+    print(np.mean(dice_core, axis=0))
+    print('mean dice enhance:')
+    print(np.mean(dice_et, axis=0))

research/cv/3dcnn/requirements.txt

+onnxruntime-gpu
+numpy
+pyyaml
+nibabel

research/cv/3dcnn/scripts/run_onnx_eval.sh

+#!/bin/bash
+# Copyright 2022 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+
+set -e
+if [ $# -ne 5 ]
+then
+    echo "=============================================================================================================="
+    echo "Please run the script as: "
+    echo "bbash run_onnx_eval.sh DATA_PATH TEST_PATH ONNX_PATH CONFIG_PATH DEVICE_ID"
+    echo "For example: bash run_onnx_eval.sh ./data ./test ./onnx default_config.ymal 0"
+    echo "It is better to use the absolute path."
+    echo "=============================================================================================================="
+exit 1
+fi
+
+DATA_PATH=$1
+TEST_PATH=$2
+ONNX_PATH=$3
+CONFIG_PATH=$4
+DEVICE_ID=$5
+
+export DATA_PATH=${DATA_PATH}
+export TEST_PATH=${TEST_PATH}
+export ONNX_PATH=${ONNX_PATH}
+export CONFIG_PATH=${CONFIG_PATH}
+export CUDA_VISIBLE_DEVICES=${DEVICE_ID}
+
+if [ ! -d "$DATA_PATH" ]; then
+        echo "dataset does not exit"
+        exit
+fi
+
+echo "eval_onnx begin."
+cd ../
+nohup python eval_onnx.py > eval_onnx.log 2>&1 \
+--batch_size=1 \
+--data_path=$DATA_PATH \
+--test_path=$TEST_PATH \
+--onnx_path=$ONNX_PATH \
+--config_path=$CONFIG_PATH \
+--device_id=$DEVICE_ID &
+echo "eval_onnx background..."