Image Classification in Nativescript via API - machine-learning

Question

I will be building an app to detect cat diseases (in their eyes) using Nativescript-Vue.
I'm searching other image classifications project and plugins in Native Script but there where little to no results. I explored google vision, Microsoft custom vision, clarify, but there were no plugins ready for Native Script.
I'm not so good yet to convert react native implementation and its possible counterpart to Native Script. Has anyone tried it?

Answer 1

You may use the MLKit from Firebase plugin which is recommended by Google for mobile apps. Google Vision also has REST apis if you like to stick with that.

Answer 2

If you want to try it but given your credit card, you may use Google Vision, Amazon Rekognition, Microsofts and other ML related in the market. But if you dont want or have a credit card number, in my case, I use https://clarifai.com/ .
For the API package,npm install --save clarifai and for the implementation, I got it with,
const Clarifai = require('clarifai')
const app = new Clarifai.App({
apiKey: 'dac044d841bb40f6bc6a79e5d1a65fcf'
})
sendToClarifai(encoded_image) {
let vm = this
app.models.predict(Clarifai.GENERAL_MODEL, { base64: encoded_image }).then(
function (response) {
// do something with response
var concepts = response['outputs'][0]['data']['concepts']
console.log('CONCEPTS', concepts)
vm.isLoading = false
let isDetectedEyeRelated = false
//loop through items
concepts.forEach(obj => {
if (['disease', 'animal', 'pet', 'cat', 'cataract', 'cataracts', 'eye', 'conjunctivitis'].includes(obj.name)) {
isDetectedEyeRelated = true
}
//more to improve here but very inaccurate
})
//show message
let message = ""
let firstQuestionArray = []
if (isDetectedEyeRelated) {
message = "Your cat might have some eye problems."
firstQuestionArray = ["Do you see redness or discharge?", "Does your cat squinting or the emergence of the third eyelid?"]
} else {
message = "Our classification model cannot identify the subject.\nCheck camera settings and proper position."
firstQuestionArray = ["Does your cat has bright, clear evenly focused eyes?", "Do you see redness or discharge?", "Does your cat squinting or the emergence of the third eyelid?"]
}
//follow up questions
action({
message: message + '\n\nBelow are follow up questions:',
cancelButtonText: "Cancel",
actions: firstQuestionArray,
cancelable: false
})
.then(result => {
if (result.includes("redness") || result.includes("emergence")) { //A
action({
message: "Please select any of these symptoms to know what seems could be the problem",
cancelButtonText: "Cancel",
actions: ["Do you see any familiar pink or reddish color, sticky eye discharge and swelling?", "Did you notice any cloudiness in their eyes, change of iris color and unusual meowing?"],
cancelable: false
})
.then(result => {
if (result.includes("reddish")) { //A
alert({
title: "Identify",
message: "This could be a sign that your cat has conjunctivitis.",
okButtonText: "See Early Treatment"
}).then(() => {
vm.setTreatmentType('eye_conjunctivitis')
vm.setTreatmentTitle('Eye Conjunctivitis Identification and Treatment')
vm.$router.replace({
path: '/pdf-treatment'
})
})
} else if (result.includes("cloudiness")) { //b
alert({
title: "Identify",
message: "This could be a sign that your cat has cataract.",
okButtonText: "See Early Treatment"
}).then(() => {
vm.setTreatmentType('eye_cataract')
vm.setTreatmentTitle('Eye Cataract Identification and Treatment')
vm.$router.replace({
path: '/pdf-treatment'
})
})
}
})
} else if (result.includes("bright")) { //b
const user_data = LS.getItem(vm.loggedUsername)
const obj = JSON.parse(user_data)
obj.progress.eye = 'good condition'
//update the user data
LS.removeItem(vm.loggedUsername)
LS.setItem(vm.loggedUsername, JSON.stringify(obj))
alert({
title: "Congratulations",
message: "Your cat's eye is in good condition.",
okButtonText: "Done"
}).then(() => {
console.log("Alert dialog closed");
})
vm.$router.push({
path: '/categories'
})
}
})
},
function (err) {
// there was an error
console.log('CLARIFAI ERROR', err)
}
)
Or something similar with my implementation.