Getting Started

Upload your data or Search the Web,
then integrate our API with a couple of lines of code.

Label Images


Select the Labels and Build a Model:




Input
Output
{
  "label": "shorts",
  "probability": 0.079877
},
{
  "label": "skirt",
  "probability": 0.002464
},
{
  "label": "jeans",
  "probability": 0.917659
}


Data Size:

Your Data:     50 Images per Label
Pretrained Model:     Pretrained on 1M+ Images

Model:

Any Image can be labeled as one of the predefined labels. Train by providing a few images of each label.

Example Use Cases:

Tag clothes, identify animals, type of trees, identify body part from an X-Ray, type of car or which season the image is taken in.

Object Detection


Select a Pretrained Model

These are models already trained on different categories

Select one of them and give it a try



OR


Train your own model

Examples could be faces, drones, logos

You will need to upload images for the category in next step



Model:

Provide an Image as an input and receive coordinates for a bounding box that surrounds the object. Train by providing a few images with corresponding bounding boxes.

Example Use Cases:

Detect where a fingerprint, bike, logo or monument is in an image.


Input
Output
{
  "label": "cat",
  "x1": 453,
  "y1": 19,
  "x2": 650,
  "y2": 250
}

Image Similarity (Coming Soon)


Model:

Provide two Images as input and receive a similarity score in response. Train by providing a few image pair with similarity scores.

Example Use Cases:

Rate how similar two products, faces, costumes or house are.


Input 1
Input 1
Output
{
    "type": "similarity",
    "score": 0.94361
},

Image Quality or Other Score (Coming Soon)


Model:

Provide an Image as an input an receive a score associated with the Image. Train by providing a few Image Score pairs

Example Use Cases:

Rate how good a photograph drawing icon or screen-shot is or assign any other custom score.


Input
Output
{
      "type": "quality",
      "score": 0.11423
}

Text Categorization


Model:

Provide a String as input and categorize it into one of the predefined categorizes. Train by providing a few Strings of each Category

Example Use Cases:

Categorize Strings into categories of reviews, types of prose or subject of text book.


Input

"This stand is about 21" tall and about 46" long it looks very nice in my living room. It was really easy to assemble it, and it has a really great color.. when it comes to align the doors or make them look straight is kind of tricky but just just have to use your screwdriver and tight it or loose it , whatever your need is, and make sure both doors are aligned the same way. If the doors are not aligned, when u close them it'll look like the doors are falling, so make sure you do it right and straight."

Output
{
  "label": "Book Review",
  "probability": 0.079877
},
{
  "label": "Movie Review",
  "probability": 0.002464
},
{
  "label": "Furniture Review",
  "probability": 0.917659
}

Text Extraction


Model:

Provide a String as input and extract predefined fields. Training by providing a few Strings with the corresponding fields.

Example Use Cases:

Extract fields from text like names, addresses, dates, times, product details or Urls.


Input

"Upcoming flight
San Francisco to Istanbul
Mar 1, 6:10 PM · Turkish 80"

Output
{
  "From": "San Francisco",
},
{
  "To": "Istanbul",
},
{
  "Date": "1st March 2017",
}

Text Similarity


Model:

Provide a Strings as input and receive a similarity score. Training by providing a few Strings pairs with scores.

Example Use Cases:

De-duplication. Compare to templates. Check if text rules are met given a format.


Input

"Crossing the Chasm, 3rd Edition: Marketing and Selling Disruptive Products to Mainstream Customers (Collins Business Essentials) Paperback"

"Crossing the Chasm: Marketing and Selling High-Tech Products to Mainstream Customers"

Output
{
      "type": "similarity",
      "score": 0.91423
}

Using NanoNets

1. Upload Data or Use Data from the Web

Upload some sample input data or Search the Web for data using our easy to use API. Specify sample output for the data. An input could be an image and output objects present in the image.

2. Instant Model

A few minutes after giving us your data you will be able to access the model. No parameter tuning, no other user input. Expected accuracies will be instantly available, in case the model isn't good enough, add more data.

3. Query our API

Integrate the model into your existing code easily using our easy to use Web API and code samples. Process data and get responses immediately. Queries can be made one at a time or in batches.

Industries

We have industry specific use cases and applications pre built to help you to get started quickly.

E - Commerce / Retail

Classification, Extraction, Similarity and Recommendation

Media / News

Image Management, Attribute Extraction and Content Filtering

Travel

Recommendation, Similarity, Extraction and Classification

Developer Friendly API and Documentation


Multiple languages supported via our easy to use API


Begin Building

How NanoNets Work

Your data + Large Pretrained Models

Recent advances in Deep Learning helped us build rich representations of data that are transferable across tasks. Using this technology we pre-train models on extremely large datasets that contain varied information. NanoNets are added to the existing model then trained on your data to solve your specific problem. Since NanoNets are smaller than traditional networks they require much less data and time to build.

Upload Data and Instantly get a ML API without tuning any parameters

NanoNets remove the hassle of tuning parameters for Machine Learning models. Upload the data wait for a couple of hours and get a model you can query over our easy to use cloud API. Data Scientists spend most time tuning parameters for feature extraction which is already baked into our pretrained models. This allows us to auto-tune parameters of NanoNets

NanoNets Constantly Improve

The world of Machine Learning is constantly evolving where feature extraction techniques improve everyday. The data available for pretraining our models also keeps growing. These two factors allow us to keep improving the pretrained model. This ensures the accuracy you receive gets better with time, without any effort. Our API allows you to provide feedback when the model makes a mistake, this is fed back into the NanoNet that improves the accuracy.

Easy to Start, convenient to Scale


Cloud based Machine Learning removes all the hassels of maintaining your own Infrastructure


Begin Building

Simple Usage Based Pricing

Free

$ 0 /
API Call
Get Started
  • 1,000 API calls
  • Fast
  • Single API Call

Basic

$ 0.001 / API Call
Get Started
  • Unlimited API calls
  • Fast
  • Single API Call

Pro

$ 0.002 / API Call
Get Started
  • Unlimited API calls
  • Lightning
  • Concurrent API Calls

Enterprise

Custom
Models
Contact Us
  • Custom Models
  • Priority
  • Concurrent API Calls

Price Calculator

API Calls

Plan:

Price:

$1

Need more help? Contact us


Let us know if you need a Custom Model, Access to Data or help Labeling your Data.


Send an email to info@nanonets.com