【PPSCI Export&Infer No. 29】 add export and inference

docs/zh/examples/topopt.md

@@ -22,6 +22,22 @@
     python topopt.py mode=eval 'EVAL.pretrained_model_path_dict={'Uniform': 'https://paddle-org.bj.bcebos.com/paddlescience/models/topopt/uniform_pretrained.pdparams', 'Poisson5': 'https://paddle-org.bj.bcebos.com/paddlescience/models/topopt/poisson5_pretrained.pdparams', 'Poisson10': 'https://paddle-org.bj.bcebos.com/paddlescience/models/topopt/poisson10_pretrained.pdparams', 'Poisson30': 'https://paddle-org.bj.bcebos.com/paddlescience/models/topopt/poisson30_pretrained.pdparams'}'
     ```
 
+=== "模型导出命令"
+
+    ``` sh
+    python topopt.py mode=export INFER.pretrained_model_name=Uniform
+    ```
+
+=== "模型推理命令"
+
+    ``` sh
+    # linux
+    wget -nc https://paddle-org.bj.bcebos.com/paddlescience/datasets/topopt/top_dataset.h5 -P ./datasets/
+    # windows
+    # curl https://paddle-org.bj.bcebos.com/paddlescience/datasets/topopt/top_dataset.h5 --output ./datasets/top_dataset.h5
+    python topopt.py mode=infer INFER.img_num=3
+    ```
+
 | 预训练模型  | 指标 |
 |:--| :--|
 | [topopt_uniform_pretrained.pdparams](https://paddle-org.bj.bcebos.com/paddlescience/models/topopt/uniform_pretrained.pdparams) | loss(sup_validator): [0.14336, 0.10211, 0.07927, 0.06433, 0.04970, 0.04612, 0.04201, 0.03566, 0.03623, 0.03314, 0.02929, 0.02857, 0.02498, 0.02517, 0.02523, 0.02618]<br>metric.Binary_Acc(sup_validator): [0.9410, 0.9673, 0.9718, 0.9727, 0.9818, 0.9824, 0.9826, 0.9845, 0.9856, 0.9892, 0.9892, 0.9907, 0.9890, 0.9916, 0.9914, 0.9922]<br>metric.IoU(sup_validator): [0.8887, 0.9367, 0.9452, 0.9468, 0.9644, 0.9655, 0.9659, 0.9695, 0.9717, 0.9787, 0.9787, 0.9816, 0.9784, 0.9835, 0.9831, 0.9845] |

examples/topopt/conf/topopt.yaml

@@ -14,8 +14,13 @@ hydra:
           - EVAL.pretrained_model_path_dict
           - EVAL.batch_size
           - EVAL.num_val_step
+          - EXPORT.pretrained_model_name
+          - INFER.pretrained_model_path_dict
+          - INFER.export_path
+          - INFER.batch_size
           - mode
           - vol_coeff
+          - log_freq
   sweep:
     # output directory for multirun
     dir: ${hydra.run.dir}
@@ -25,6 +30,7 @@ hydra:
 mode: train # running mode: train/eval
 seed: 42
 output_dir: ${hydra:run.dir}
+log_freq: 20
 
 # set default cases parameters
 CASE_PARAM: [[Poisson, 5], [Poisson, 10], [Poisson, 30], [Uniform, null]]
@@ -57,3 +63,28 @@ EVAL:
   pretrained_model_path_dict: null # a dict: {casename1:path1, casename2:path2, casename3:path3, casename4:path4}
   num_val_step: 10 # the number of iteration for each evaluation case
   batch_size: 16
+
+# inference settings
+INFER:
+  pretrained_model_name: null # a string, indicating which model you want to export. Support [Uniform, Poisson5, Poisson10, Poisson30].
+  pretrained_model_path_dict: {'Uniform': 'https://paddle-org.bj.bcebos.com/paddlescience/models/topopt/uniform_pretrained.pdparams', 'Poisson5': 'https://paddle-org.bj.bcebos.com/paddlescience/models/topopt/poisson5_pretrained.pdparams', 'Poisson10': 'https://paddle-org.bj.bcebos.com/paddlescience/models/topopt/poisson10_pretrained.pdparams', 'Poisson30': 'https://paddle-org.bj.bcebos.com/paddlescience/models/topopt/poisson30_pretrained.pdparams'}
+  export_path: ./inference/topopt
+  pdmodel_path: ${INFER.export_path}.pdmodel
+  pdpiparams_path: ${INFER.export_path}.pdiparams
+  device: gpu
+  engine: native
+  precision: fp32
+  onnx_path: null
+  ir_optim: true
+  min_subgraph_size: 30
+  gpu_mem: 4000
+  gpu_id: 0
+  max_batch_size: 1024
+  num_cpu_threads: 10
+  batch_size: 4
+  sampler_key: Fixed # a string, indicating the sampling method. Support [Fixed, Uniform, Poisson].
+  sampler_num: 8 # a integer number, indicating the sampling rate of the sampling method, supported when `sampler_key` is Fixed or Poisson.
+  img_num: 4
+  res_img_figsize: null
+  save_res_path: ./inference/predicted
+  save_npy: false

examples/topopt/topopt.py

@@ -17,13 +17,16 @@
 import functions as func_module
 import h5py
 import hydra
+import matplotlib.pyplot as plt
 import numpy as np
 import paddle
 from omegaconf import DictConfig
 from paddle import nn
+from paddle.static import InputSpec
 from topoptmodel import TopOptNN
 
 import ppsci
+from deploy.python_infer import pinn_predictor
 from ppsci.utils import logger
 
 
@@ -120,7 +123,7 @@ def evaluate(cfg: DictConfig):
 
     # fixed iteration stop times for evaluation
     iterations_stop_times = range(5, 85, 5)
-    model = TopOptNN()
+    model = TopOptNN(**cfg.MODEL)
 
     # evaluation for 4 cases
     acc_results_summary = {}
@@ -317,14 +320,120 @@ def val_metric(output_dict, label_dict, weight_dict=None):
     return {"Binary_Acc": acc, "IoU": iou}
 
 
+# export model
+def export(cfg: DictConfig):
+    # set model
+    model = TopOptNN(**cfg.MODEL)
+
+    # initialize solver
+    solver = ppsci.solver.Solver(
+        model,
+        eval_with_no_grad=True,
+        pretrained_model_path=cfg.INFER.pretrained_model_path_dict[
+            cfg.EXPORT.pretrained_model_name
+        ],
+    )
+
+    # export model
+    input_spec = [{"input": InputSpec([None, 2, 40, 40], "float32", name="input")}]
+
+    solver.export(input_spec, cfg.INFER.export_path)
+
+
+# model inference
+def inference(cfg: DictConfig):
+    # read h5 data
+    h5data = h5py.File(cfg.DATA_PATH, "r")
+    data_iters = np.array(h5data["iters"])
+    data_targets = np.array(h5data["targets"])
+    idx = np.random.choice(len(data_iters), cfg.INFER.img_num, False)
+    data_iters = data_iters[idx]
+    data_targets = data_targets[idx]
+
+    sampler = func_module.generate_sampler(cfg.INFER.sampler_key, cfg.INFER.sampler_num)
+    data_iters = channel_sampling(sampler, data_iters)
+
+    predictor = pinn_predictor.PINNPredictor(cfg)
+
+    input_dict = {"input": data_iters}
+    output_dict = predictor.predict(input_dict, cfg.INFER.batch_size)
+
+    # mapping data to output_key
+    output_dict = {
+        store_key: output_dict[infer_key]
+        for store_key, infer_key in zip({"output"}, output_dict.keys())
+    }
+
+    save_topopt_img(
+        input_dict,
+        output_dict,
+        data_iters,
+        cfg.INFER.save_res_path,
+        cfg.INFER.res_img_figsize,
+        cfg.INFER.save_npy,
+    )
+
+
+# used for inference
+def channel_sampling(sampler, input):
+    SIMP_initial_iter_time = sampler()
+    input_channel_k = input[:, SIMP_initial_iter_time, :, :]
+    input_channel_k_minus_1 = input[:, SIMP_initial_iter_time - 1, :, :]
+    input = np.stack(
+        (input_channel_k, input_channel_k - input_channel_k_minus_1), axis=1
+    )
+    return input
+
+
+# used for inference
+def save_topopt_img(
+    input_dict, output_dict, ground_truth, res_path, figsize=None, npy=False
+):
+
+    input = input_dict["input"]
+    output = output_dict["output"]
+    for i in range(len(input)):
+        plt.figure(figsize=figsize)
+        plt.subplot(1, 4, 1)
+        plt.axis("off")
+        plt.imshow(input[i][0], cmap="gray")
+        plt.title("Input Image")
+        plt.subplot(1, 4, 2)
+        plt.axis("off")
+        plt.imshow(input[i][1], cmap="gray")
+        plt.title("Input Gradient")
+        plt.subplot(1, 4, 3)
+        plt.axis("off")
+        plt.imshow(np.round(output[i][0]), cmap="gray")
+        print(output[i])
+        plt.title("Prediction")
+        plt.subplot(1, 4, 4)
+        plt.axis("off")
+        plt.imshow(np.round(ground_truth[i][0]), cmap="gray")
+        print(ground_truth[i])
+        plt.title("Ground Truth")
+        plt.show()
+        plt.savefig(osp.join(res_path, "Prediction_" + str(i) + ".png"))
+        plt.close()
+        if npy:
+            with open(osp(res_path, "Prediction_" + str(i) + ".npy"), "wb") as f:
+                np.save(f, output[i])
+
+
 @hydra.main(version_base=None, config_path="./conf", config_name="topopt.yaml")
 def main(cfg: DictConfig):
     if cfg.mode == "train":
         train(cfg)
     elif cfg.mode == "eval":
         evaluate(cfg)
+    elif cfg.mode == "export":
+        export(cfg)
+    elif cfg.mode == "infer":
+        inference(cfg)
     else:
-        raise ValueError(f"cfg.mode should in ['train', 'eval'], but got '{cfg.mode}'")
+        raise ValueError(
+            f"cfg.mode should in ['train', 'eval', 'export', 'infer'], but got '{cfg.mode}'"
+        )
 
 
 if __name__ == "__main__":

examples/topopt/topoptmodel.py

@@ -32,7 +32,10 @@ class TopOptNN(ppsci.arch.UNetEx):
         kernel_size (int, optional): Size of kernel of convolution layer. Defaults to 3.
         filters (Tuple[int, ...], optional): Number of filters. Defaults to (16, 32, 64).
         layers (int, optional): Number of encoders or decoders. Defaults to 3.
-        channel_sampler (callable): The sampling function for the initial iteration time (corresponding to the channel number of the input) of SIMP algorithm.
+        channel_sampler (callable, optional): The sampling function for the initial iteration time
+                (corresponding to the channel number of the input) of SIMP algorithm. The default value
+                is None, when it is None, input for the forward method should be sampled and prepared
+                with the shape of [batch, 2, height, width] before passing to forward method.
         weight_norm (bool, optional): Whether use weight normalization layer. Defaults to True.
         batch_norm (bool, optional): Whether add batch normalization layer. Defaults to True.
         activation (Type[nn.Layer], optional): Name of activation function. Defaults to nn.ReLU.
@@ -51,7 +54,7 @@ def __init__(
         kernel_size=3,
         filters=(16, 32, 64),
         layers=2,
-        channel_sampler=lambda: 1,
+        channel_sampler=None,
         weight_norm=False,
         batch_norm=False,
         activation=nn.ReLU,
@@ -124,15 +127,17 @@ def __init__(
         )
 
     def forward(self, x):
-        SIMP_initial_iter_time = self.channel_sampler()  # channel k
-        input_channel_k = x[self.input_keys[0]][:, SIMP_initial_iter_time, :, :]
-        input_channel_k_minus_1 = x[self.input_keys[0]][
-            :, SIMP_initial_iter_time - 1, :, :
-        ]
-        x = paddle.stack(
-            (input_channel_k, input_channel_k - input_channel_k_minus_1), axis=1
-        )
-
+        if self.channel_sampler is not None:
+            SIMP_initial_iter_time = self.channel_sampler()  # channel k
+            input_channel_k = x[self.input_keys[0]][:, SIMP_initial_iter_time, :, :]
+            input_channel_k_minus_1 = x[self.input_keys[0]][
+                :, SIMP_initial_iter_time - 1, :, :
+            ]
+            x = paddle.stack(
+                (input_channel_k, input_channel_k - input_channel_k_minus_1), axis=1
+            )
+        else:
+            x = x[self.input_keys[0]]
         # encode
         upsampling_size = []
         skip_connection = []