React Native加载PyTorch模型遇Format error报错求助

错误信息 #

Possible Unhandled Promise Rejection (id: 1):
Object {
  "message": "Format error
Exception raised from _load_for_mobile at /data/users/atalman/pytorch/torch/csrc/jit/mobile/import.cpp:623 (most recent call first):
(no backtrace available)",
}

背景 #

开发React Native目标检测应用时遇到上述错误，严格遵循playtorch.dev 0.2.4版本的目标检测教程开发，已尝试重新安装PyTorch但问题未解决。

项目代码 #

App.js #

import {StyleSheet, Text, View} from 'react-native';
import React from 'react';
import {
  Camera,
  MobileModel,
  torch,
  torchvision,
  media,
} from 'react-native-pytorch-core';
let model = null;
const T = torchvision.transforms;
const App = () => {
  async function handleImage(image) {
    console.log('Image Taken!!');
    const width = image.getWidth();
    const height = image.getHeight();
    // 3.ii. Convert image to blob, which is a byte representation of the image
    // in the format height (H), width (W), and channels (C), or HWC for short
    const blob = media.toBlob(image);
    // 3.iii. Get a tensor from image the blob and also define in what format
    // the image blob is.
    let tensor = torch.fromBlob(blob, [height, width, 3]);
    // 3.iv. Rearrange the tensor shape to be [CHW]
    tensor = tensor.permute([2, 0, 1]);
    // 3.v. Divide the tensor values by 255 to get values between [0, 1]
    tensor = tensor.div(255);
    // 3.vi. Crop the image in the center to be a squared image
    const centerCrop = T.centerCrop(Math.min(width, height));
    tensor = centerCrop(tensor);
    // 3.vii. Resize the image tensor to 3 x 224 x 224
    const resize = T.resize(224);
    tensor = resize(tensor);
    // 3.viii. Normalize the tensor image with mean and standard deviation
    const normalize = T.normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]);
    tensor = normalize(tensor);
    // 3.ix. Unsqueeze adds 1 leading dimension to the tensor
    tensor = tensor.unsqueeze(0);
    // console.log(tensor);
    // 3.x. Return the tensor shape [1, 3, 224, 224]
    const result = tensor.shape;
    console.log('result:', result);
    if (model == null) {
      console.log('Loading model...');
      const filePath = await MobileModel.download('detr_resnet50.ptl');
      model = await torch.jit._loadForMobile(filePath);
      console.log('Model successfully loaded');
    }
    console.log('Forward propogation !!');
    const output = await model.forward(tensor);
    console.log(output);
  }
  return (
    <View style={styles.container}>
      <Text style={styles.label}>Class: </Text>
      <Camera style={styles.camera} onCapture={handleImage} />
    </View>
  );
};

export default App;

const styles = StyleSheet.create({
  container: {
    flexGrow: 1,
    backgroundColor: '#ffff',
    padding: 20,
    alignItems: 'center',
  },
  label: {
    marginBottom: 10,
    color: 'black',
    fontSize: 20,
  },
  camera: {
    flexGrow: 1,
    width: '100%',
    marginTop: 70,
  },
});

项目依赖配置（devDependencies） #

{
  "@babel/core": "^7.12.9",
  "@babel/runtime": "^7.12.5",
  "@react-native-community/eslint-config": "^2.0.0",
  "babel-jest": "^26.6.3",
  "eslint": "^7.32.0",
  "jest": "^26.6.3",
  "metro-react-native-babel-preset": "0.72.3",
  "react-test-renderer": "18.1.0"
}

排查与解决方案 #

• 验证模型完整性：检查下载的detr_resnet50.ptl文件是否完整，是否为PyTorch Mobile兼容格式。可手动重新下载模型，确认文件无损坏或中断情况。
• 替换模型加载方法：将torch.jit._loadForMobile替换为公开APItorch.jit.load，部分版本中_loadForMobile为内部方法，容易出现兼容性问题。
• 匹配依赖版本：确认react-native-pytorch-core版本与教程的0.2.4一致，执行npm list react-native-pytorch-core查看当前版本，若不符则安装指定版本：npm install react-native-pytorch-core@0.2.4。
• 调整模型加载时机：将模型加载逻辑移至组件挂载阶段，避免在相机捕获回调中异步加载，减少竞争问题：

  React.useEffect(() => {
    async function loadModel() {
      console.log('Loading model...');
      const filePath = await MobileModel.download('detr_resnet50.ptl');
      model = await torch.jit.load(filePath);
      console.log('Model successfully loaded');
    }
    loadModel();
  }, []);
  

• 清理项目缓存：执行以下命令清除缓存并重新安装依赖：

  npm cache clean --force
  rm -rf node_modules package-lock.json
  npm install
  npx react-native start --reset-cache
  

• 检查存储权限：Android端需在AndroidManifest.xml中添加存储读写权限，适配对应系统版本的权限要求。

内容的提问来源于stack exchange，提问作者shubham Sahu