Blob and Base64 images to Tensor

Given the same .jpg image, HTML5 commonly accepts these types of <img /> tags:

<img src="https://domain.com/path/to/my/image.jpg" />
<img src="data:image/jpg;base64,/dh7hfhwehfw9hfw...wh9fh8w9hf" />
<img src="blob:https://domain.com/b3872f9f-acb7-49d9-a26d-4854d72c9939" />

If I am correct, tfjs will only load images using tf.fromPixels(input) which, as its name implies, loads from pixels (= the first type of src). However, I could not find anything like tf.fromBase64(input) nor tf.fromBlob(input) to convert other formats to tensors. More generally, is there any way to process Blobs and/or base64 images through tfjs ? Thanks !

[0x00b1]

fromPixels accepting a base64 string would be super useful since the HTMLImageElement (or similar) dance is especially cumbersome in our application where images are stored in an IndexedDB database and rarely outputted to the DOM.

[rthadur]

Automatically closing due to lack of recent activity. Please update the issue when new information becomes available, and we will reopen the issue. Thanks!