qing
d17677bade
- export_onnx: opset 17→18，dynamic_axes 改 dynamic_shapes 消除弃用警告 - predict: action_stats 改为 torch tensor，消除 NumPy 2.0 __array_wrap__ 警告 - train/pretrain: 每 epoch 保存 last.pt，新增 --resume 断点续训练 - model: MobileNetEncoder 下载失败时回退到已缓存权重 - _set.sh: 安装时预缓存 MobileNetV2 权重，失败给出明确指引 - test: 新增 TestResumeCheckpoint（2 用例），总计 58 测试全绿
59 lines
2.0 KiB
Python
59 lines
2.0 KiB
Python
import argparse
|
||
import torch
|
||
from torch.export import Dim
|
||
from config import parse_checkpoint_config
|
||
from model import ActionPredictor
|
||
|
||
|
||
def main():
|
||
parser = argparse.ArgumentParser(description="Export actpred model to ONNX")
|
||
parser.add_argument("--model", required=True, help="checkpoint path (best_inference.pt)")
|
||
parser.add_argument("--output", default="model.onnx", help="output ONNX file path")
|
||
args = parser.parse_args()
|
||
|
||
checkpoint = torch.load(args.model, map_location="cpu", weights_only=True)
|
||
cfg = parse_checkpoint_config(checkpoint["config"])
|
||
data_cfg = cfg.data
|
||
model_cfg = cfg.model
|
||
|
||
model = ActionPredictor(model_cfg, data_cfg, pretrained=False)
|
||
model.load_state_dict(checkpoint["model_state_dict"])
|
||
model.eval()
|
||
|
||
dummy_images = torch.randn(
|
||
1, data_cfg.frame_len, 3, data_cfg.image_size, data_cfg.image_size)
|
||
dummy_history = torch.randn(1, data_cfg.history_len, data_cfg.action_dim)
|
||
|
||
batch = Dim("batch")
|
||
torch.onnx.export(
|
||
model, (dummy_images, dummy_history),
|
||
args.output,
|
||
input_names=["images", "action_history"],
|
||
output_names=["predicted_action"],
|
||
dynamic_shapes={
|
||
"images": {0: batch},
|
||
"action_history": {0: batch},
|
||
},
|
||
opset_version=18,
|
||
)
|
||
print(f"ONNX model exported to: {args.output}")
|
||
|
||
# 可选：验证导出的 ONNX 模型一致性
|
||
try:
|
||
import onnxruntime as ort
|
||
import numpy as np
|
||
sess = ort.InferenceSession(args.output)
|
||
img_np = dummy_images.numpy()
|
||
hist_np = dummy_history.numpy()
|
||
ort_out = sess.run(None, {"images": img_np, "action_history": hist_np})[0]
|
||
with torch.no_grad():
|
||
torch_out = model(dummy_images, dummy_history).numpy()
|
||
max_diff = np.abs(ort_out - torch_out).max()
|
||
print(f"ONNX 一致性验证通过（最大差异: {max_diff:.6e}）")
|
||
except ImportError:
|
||
print("提示: 安装 onnxruntime 可验证导出模型一致性 (pip install onnxruntime)")
|
||
|
||
|
||
if __name__ == "__main__":
|
||
main()
|