Deploying using api is giving 500 eroor

my project type: object detection
i trained my model and when deploying the image is not being taken it is giving error saying
‘cannot identify image file <_io.BytesIO object at 0x7f2f3c3d1ea0>’. I am doing this in javascript.
Any help is appreciated.Thnak you

Can you provide more information? What does your code look like? @karthik_Mutyala

I have the same problem.

Hi @jackey - what code are you using?

I’ll be unable to debug unless I have an example

I am using js

Robo Flow Engineers please help! I am facing the exact same issue. Followed the API documentation properly. Below is my code snippet. I am using the language Dart. For mobile app development, with the Flutter framework. I am also getting the same error all the time, no matter what I do.

@jackey @zong_wei

Interesting - I’m having trouble reproducing this. Are hosted images the only images you are testing with, or have you both tried with local images, too?

Additionally, what are the image sizes you’re testing with? (In Megabytes, and the width/height in pixels)

The image size is 1.12MB with a dimension of 1920 x 1280. (local image)
It did not work for hosted images too. initially there was a length error with which I added some padding as you can see in the code snippet above. The base64 bytes are all good I have checked.

Hello, even the simplest post request using postman with a base64 encoded image string fails to work. This is most likely a bug on roboflow end, I have verified my API key and credentials, and the base64 image string is verified to be non corrupt.

Is it possible for me to download my model weights instead of inferencing over API call. There does not seem to have an option to download my trained weights. Thank you