Salesforce Einstein Vision: How it works?

Salesforce Einstein: 
Salesforce Einstein is a powerful APIs with great machine learning algorithms. Salesforce has two versions of these APIs. V2 is in beta. Here are they
1. Salesforce Einstein Version-1 (v1) comes with Image classification APIs
2. Salesforce Einstein Version-2 (v2) comes with Natural Language Processing (Get the sentiment of the sentence and Intent processing). + V1

Salesforce Einstein Vision is all about following five terms:
1. Dataset
2. Labels
3. Training
4. Models
5. Prediction

Let’s take an example of Image Classifier with Salesforce Einstein Vision. We are going to predict about fruits. So let’s assume we have 3 fruits here with their few images (Mango, Banana, Grapes).

1. Dataset and Labels:

Here Fruit is the dataset and labels are the names of the fruits.

Note: Related APIs->https://api.einstein.ai/v2/language/datasets/upload, https://api.einstein.ai/v1/vision/datasets/upload/sync

2. Creating and Training Model:

Now next process is Train the data and creates model from our dataset. We will pass this dataset to Train API. Here we can decide how much data we need to use as Training data and Testing data. Train API have trainParams={“trainSplitRatio”: 0.n} flag in the API. Here n is Trading data. If you write 0.7 that means, 70% data will be used as training data and 30% will be used as test data.

Note: Related APIs->https://api.einstein.ai/v1/vision/train

3. Prediction:

Now it’s time to rock with Salesforce Einstein. We will send an image of banana to Salesforce Einstein with our model of fruits. It will return the predicted output as Banana.

Note: Related APIs->https://api.einstein.ai/v1/vision/predict

I hope this will give you some basic startup with Salesforce Einstein Vision. Read detail APIs here. https://metamind.readme.io/v1/docs/

Cheers!!