Home IT技术 如何在网页上运行我的model.json文件来测试预测？

如何在网页上运行我的model.json文件来测试预测？

IT技术 · 2025年5月30日 · 0 Comment

我有一个Tensorflow.js的model.json文件和相关的bin文件放在文件夹里，并且我已经创建了一个html文件。我希望在html文件中测试我的预测并获得反馈，但当我在服务器上运行它时却不工作，有什么想法吗？

<html><head><script src="https://cdn.jsdelivr.net/npm/@tensorflow/tfjs@latest"></script><script>    async function run(){        const MODEL_URL = 'http://127.0.0.1:8887/model.json';        const model = await tf.loadLayersModel(MODEL_URL);        console.log(model.summary());        const result = model.predict(input);        alert(result);    }    run();</script><body></body></head>

在开发者工具中显示如下信息：

DevTools failed to load SourceMap: Could not load content for https://cdn.jsdelivr.net/npm/@tensorflow/tf.min.js.map: HTTP error: status code 404, net::ERR_HTTP_RESPONSE_CODE_FAILURE

它还显示配置格式不正确：

tfjs@latest:17 Uncaught (in promise) Error: layer: Improper config format: {"node":[{"name":"eedde323-f0a1-4978-8c57-3ae553bfb873/tf_op_layer_3ff186c1-fe28-4442-abe1-44f7ed005781/19c66cd4-e1d6-49f4-80ab-3aa348c5b264/ff9b2ff4-4799-495e-8e91-a6fed91e9315/GatherV2/3ff186c1-fe28-4442-abe1-44f7ed005781/19c66cd4-e1d6-49f4-80ab-3aa348c5b264/ff9b2ff4-4799-495e-8e91-a6fed91e9315/GatherV2/params","op":"Const","attr":{"value":{"tensor":{"dtype":"DTSTRING","tensorShape":{"dim":[{"size":"2"}]}}},"dtype":{"type":"DTSTRING"}}},{"name":"conv5_block1_0_conv/kernel","op":"Const","attr":{"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"1"},{"size":"1"},{"size":"1024"},{"size":"2048"}]}}},"dtype":{"type":"DTFLOAT"}}},{"name":"conv5_block1_0_conv/bias","op":"Const","attr":{"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"2048"}]}}},"dtype":{"type":"DTFLOAT"}}},{"name":"conv4_block1_0_conv/kernel","op":"Const","attr":{"dtype":{"type":"DTFLOAT"},"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"1"},{"size":"1"},{"size":"512"},{"size":"1024"}]}}}}},{"name":"conv4_block1_0_conv/bias","op":"Const","attr":{"dtype":{"type":"DTFLOAT"},"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"1024"}]}}}}},{"name":"conv3_block1_0_conv/kernel","op":"Const","attr":{"dtype":{"type":"DTFLOAT"},"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"1"},{"size":"1"},{"size":"256"},{"size":"512"}]}}}}},{"name":"conv3_block1_0_conv/bias","op":"Const","attr":{"dtype":{"type":"DTFLOAT"},"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"512"}]}}}}},{"name":"conv2_block1_0_conv/kernel","op":"Const","attr":{"dtype":{"type":"DTFLOAT"},"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"1"},{"size":"1"},{"size":"64"},{"size":"256"}]}}}}},{"name":"conv2_block1_0_conv/bias","op":"Const","attr":{"dtype":{"type":"DTFLOAT"},"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"256"}]}}}}},{"name":"Image","op":"Placeholder","attr":{"dtype":{"type":"DTFLOAT"},"shape":{"shape":{"dim":[{"size":"-1"},{"size":"224"},{"size":"224"},{"size":"3"}]}}}},{"name":"saved_resnet_50_model/tf_op_layer_Mul/Mul/y","op":"Const","attr":{"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{}}},"dtype":{"type":"DTFLOAT"}}},{"name":"saved_resnet_50_model/tf_op_layer_Sub/Sub/y","op":"Const","attr":{"dtype":{"type":"DTFLOAT"},"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{}}}}},{"name":"saved_resnet_50_model/resnet50v2/conv1_pad/Pad/paddings","op":"Const","attr":{"value":{"tensor":{"dtype":"DTINT32","tensorShape":{"dim":[{"size":"4"},{"size":"2"}]}}},"dtype":{"type":"DTINT32"}}},{"name":"conv1_conv/kernel","op":"Const","attr":{"dtype":{"type":"DTFLOAT"},"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"7"},{"size":"7"},{"size":"3"},{"size":"64"}]}}}}},{"name":"conv1_conv/bias","op":"Const","attr":{"dtype":{"type":"DTFLOAT"},"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"64"}]}}}}},{"name":"saved_resnet_50_model/resnet50v2/pool1_pad/Pad/paddings","op":"Const","attr":{"value":{"tensor":{"dtype":"DTINT32","tensorShape":{"dim":[{"size":"4"},{"size":"2"}]}}},"dtype":{"type":"DTINT32"}}},{"name":"saved_resnet_50_model/resnet50v2/conv2_block1_2_pad/Pad/paddings","op":"Const","attr":{"dtype":{"type":"DTINT32"},"value":{"tensor":{"dtype":"DTINT32","tensorShape":{"dim":[{"size":"4"},{"size":"2"}]}}}}},{"name":"conv2_block1_3_conv/kernel","op":"Const","attr":{"dtype":{"type":"DTFLOAT"},"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"1"},{"size":"1"},{"size":"64"},{"size":"256"}]}}}}},{"name":"conv2_block1_3_conv/bias","op":"Const","attr":{"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"256"}]}}},"dtype":{"type":"DTFLOAT"}}},{"name":"saved_resnet_50_model/resnet50v2/conv2_block2_2_pad/Pad/paddings","op":"Const","attr":{"dtype":{"type":"DTINT32"},"value":{"tensor":{"dtype":"DTINT32","tensorShape":{"dim":[{"size":"4"},{"size":"2"}]}}}}},{"name":"conv2_block2_3_conv/kernel","op":"Const","attr":{"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"1"},{"size":"1"},{"size":"64"},{"size":"256"}]}}},"dtype":{"type":"DTFLOAT"}}},{"name":"conv2_block2_3_conv/bias","op":"Const","attr":{"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"256"}]}}},"dtype":{"type":"DTFLOAT"}}},{"name":"saved_resnet_50_model/resnet50v2/conv2_block3_2_pad/Pad/paddings","op":"Const","attr":{"value":{"tensor":{"dtype":"DTINT32","tensorShape":{"dim":[{"size":"4"},{"size":"2"}]}}},"dtype":{"type":"DTINT32"}}},{"name":"conv2_block3_3_conv/kernel","op":"Const","attr":{"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"1"},{"size":"1"},{"size":"64"},{"size":"256"}]}}},"dtype":{"type":"DTFLOAT"}}},{"name":"conv2_block3_3_conv/bias","op":"Const","attr":{"value":{"tensor":{"dtype":"DTFLOAT","tensorShape":{"dim":[{"size":"256"}]}}},"dtype":{"type":"DTFLOAT"}}},{"name":"saved_resnet_50_model/resnet50v2/conv3_block1_2_patfjs@latest:17)    at tA (tfjs@latest:17)    at tfjs@latest:17    at u (tfjs@latest:17)    at Generator._invoke (tfjs@latest:17)    at Generator.forEach.e.<computed> [as next] (tfjs@latest:17)    at Um (tfjs@latest:17)    at o (tfjs@latest:17)Jm @ tfjs@latest:17r @ tfjs@latest:17t @ tfjs@latest:17eE @ tfjs@latest:17tA @ tfjs@latest:17(anonymous) @ tfjs@latest:17u @ tfjs@latest:17(anonymous) @ tfjs@latest:17forEach.e.<computed> @ tfjs@latest:17Um @ tfjs@latest:17o @ tfjs@latest:17

回答：

我可以通过以下步骤避免模型的错误：

  1. 以.h5格式保存Tensorflow模型：

    model.save('sampleModel.h5')

  2. 将.h5文件转换为tfjs_graph_model格式以便在tensorflow.js中使用：

    tensorflowjs_converter --input_format keras --output_format tfjs_graph_model sampleModel.h5 sampleModel

  3. 使用loadGraphModel()而不是loadLayerModel()来加载模型：

    <html><head><script src="https://cdn.jsdelivr.net/npm/@tensorflow/[email protected]/dist/tf.min.js"></script></head><body><script>tf.loadGraphModel('sampleModel/model.json')</script></body></html>

欲了解更多信息，请查看这里

为了避免导入tensorflow.js库时出现错误，请使用此链接："https://cdn.jsdelivr.net/npm/@tensorflow/tfjs"，而不是此链接："https://cdn.jsdelivr.net/npm/@tensorflow/tfjs@latest"。

相关文章：

  1. tensorflowjs 如何在 CNN 预测中获取内部层的输出
  2. 检查输入时出错：期望dense_Dense5_input具有4个维度，但得到的数组形状为5,2,5
  3. TensorFlow.js中用于颜色预测的最佳模型类型？
  4. Handpose tfjs 错误 – 注册表中未找到后端
  5. 如何在使用Flask部署模型时预处理新数据集
  6. 使用TensorFlow JS的OpenAI GPT-2模型
  7. Tensorflowjs 预测类别
  8. 大小(27900)必须与形状1,224,224,3的乘积相匹配
  9. TensorFlow JS – 保存模型时一并保存最大最小值，并在预测数据时加载这些值
  10. 未知正则化器：TensorFlow.js中的L2
xiaolong

Related Posts

IT技术

使用LSTM在Python中预测未来值

· 2025年6月1日 · 0 Comment

这段代码可以预测指定股票的当前日期之前的值，但不能预测…

IT技术

如何在gensim的word2vec模型中查找双词组的相似性

· 2025年6月1日 · 0 Comment

我有一个word2vec模型，假设我使用的是googl…

IT技术

dask_xgboost.predict 可以工作但无法显示 – 数据必须是一维的

· 2025年6月1日 · 0 Comment

我试图使用 XGBoost 创建模型。 看起来我成功地…

IT技术

ML Tuning – Cross Validation in Spark

· 2025年6月1日 · 0 Comment

我在https://spark.apache.org/…

IT技术

如何在React JS中使用fetch从REST API获取预测

· 2025年6月1日 · 0 Comment

我正在开发一个应用程序，其中Flask REST AP…

IT技术

如何分析ML.NET中多类分类预测得分数组？

· 2025年6月1日 · 0 Comment

我在ML.NET中创建了一个多类分类项目。该项目可以对…

发表回复

您的邮箱地址不会被公开。 必填项已用 * 标注