cloud-annotations / javascript-sdk

Use custom trained Cloud Annotations models with TensorFlow.js

Cloud Annotations

Important Note: Due to unforseen circumstances, the hosted version of Cloud Annotations (cloud.annotations.ai) is no longer available.

For a similar hosted experience, but focused on education/learning, see Computer Vision Learning Studio.

For instructions on how to run Cloud Annotations using the local file system, see the experimental Iris project.

javascript-sdk's Issues

predictions.map is not a function

when i test my model
i got this error
error image

https://pasteboard.co/Ja7088N.png

please help me

error

Hello @bourdakos1
i have two problem
1- when i annotate new object in my unlabeled images , cloud not save my draw and reset all image to undraw state

2- when i click train button
it gives red error status without any message

error when import cloud-annotations

i make simple electron app and i install this package
https://www.npmjs.com/package/@cloud-annotations/models

and when i import it like this code i got this error ,

Uncaught SyntaxError: Unexpected token import C:\Users\mas\Desktop\models\electron\WebChimera-Electron-quick-start-master\node_modules@cloud-annotation\models\src\index.js:1

the image
https://pasteboard.co/JagElbW.png

i can't understand what is for ,
i installed tensorflowjs these version but get error .
"@tensorflow/tfjs": "^1.7.4",
"@tensorflow/tfjs-node": "^1.7.4"

load cloud-annotations script locally

Hello
is it possible load this script

<script src="https://cdn.jsdelivr.net/npm/@cloud-annotations/models"></script>

locally (offline) without internet ???
beacause it needs tensorflow js and i'm not sure that it works offline

if this is possible please explain what file need to download and how use all of them??

USE JAVASCRIPT-SDK IN NODE-RED

is it possible use object detection in node-red ????

can you send source code for check ?

increase accuracy

hello @bourdakos1

i want to make simple ocr project
do i make ocr system with cloud annotation ???

how many images is require for every character (ocr) with high accuracy ??
is there config for increase accuracy in cloud annotation ???

object detection on object HTMLImageElement

i test image by this code but my model not detect anything

const model22mas = useModel222(process.env.PUBLIC_URL + '/model_web')
const frame1 = document.getElementById('img1')
if( frame1== undefined){
console.log('frame11 error undefined')
}else{
const predictions = model22mas.detect( frame1 )
console.log("predictions " + predictions )

Expose setBackend endpoint

Would it be possible to expose a backend setter in this SDK, or is this library built around the CPU backend?

I would like to experiment with different backends and see their performance on mobile devices. According to the docs WebGL should be most efficient of the alternatives.

cacti train fails with WML error

When training, cacli fails with the below error

error Unable to get /v3/wml_instances/33b5e230-4230-4516-8d87-735d71b70f34 [400]: {"trace":"b401ace168081dfe7f3e517551f7b550","errors":[{"code":"invalid_request","message":"Instance in v2 plan should use v4 API."}]}