CODEWITHSUNDEEP
google-chrometensorflow.js
Labeo
Labeo

Reputation: 6339

Error while trying to load tensorflowJS model from local, Fetch API cannot load downloads://model. URL scheme must be"https" for CORS request error

I am trying to load a tensorflow js model that is saved in downloads directory as mentioned in the tutorials of tensorflowjs. But I am facing cors error please find the image below.

Code:

<html>

<head>


<script src="https://cdn.jsdelivr.net/npm/@tensorflow/[email protected]/dist/tf.min.js"></script>

<script>

async function app() {

const t = await tf.loadLayersModel('downloads://model');
console.log("done");
console.log(t);
}

app();


</script>
<script>

</script>

</head>

<body>

</body>
</html>

enter image description here

Any pointers on how to resolve this.

Tried disabling cors for chrome but still didnt work.

Upvotes: 3

Views: 2853

Answers (3)

redjack001
redjack001

Reputation: 51

I have the similar issue when I was doing my Coursera assignment.

The problem was not exactly with the tensorflow.js.

I solved the problem by pointing the CSV directly to github URL like

https://raw.githubusercontent.com/...../...../testingdata.csv

instead of putting them in my own laptop directories.

Then it should work.

Upvotes: 0

edkeveked
edkeveked

Reputation: 18371

Fetch API cannot load downloads://model. URL scheme must be“https” for CORS request error

The error says it all. According to the official docs, here is what we can read for the url parameter

A string path to the ModelAndWeightsConfig JSON describing the model in the canonical TensorFlow.js format. For file:// (tfjs-node-only), http:// and https:// schemas, the path can be either absolute or relative.

download does not mean anything, at least not when loading a model. It is instead used when downloading the model to the file system.

Therefore, if you're loading your model in the browser, consider first serving your model by a server and use a syntax like

const t = await tf.loadLayersModel('http://path/to/downloads/model.json'); // it can also be https if the server has a ssl certificate

If working in a nodejs server environment, then the file can be accessed directly using file identifier

const t = await tf.loadLayersModel('file://path/to/downloads/model.json');

Upvotes: 0

Kai Sasaki
Kai Sasaki

Reputation: 687

The protocol you specified looks invalid. You can specify file:// or just omitting it should work. And also you need to specify the path to model.json file created by tfjs-converter. So overall, the code to load the model may look like this.

const t = await tf.loadLayersModel('file:///path/to/downloads/model.json');

Upvotes: 0

Related Questions