Pre-trained machine learning models available in AWS Marketplace

Build trained AI models and ML powered solutions faster.

What are pre-trained machine learning models?

Pre-trained Machine Learning (ML) models are ready-to-use models that can be quickly deployed on Amazon SageMaker, a fully managed cloud machine learning platform. By pre-training the ML models for you, solutions in AWS Marketplace take care of the heavy lifting, helping you deliver AI and ML powered features faster and at a lower cost.

Benefits of using pre-trained models in AWS Marketplace

Faster AI/ML solution development

▶ AWS Marketplace offers hundreds of pre-trained models for tasks like object detection, buyer propensity, natural language processing, data extraction, and feature engineering.
 
▶ Pre-trained models are ready to use and can be integrated with your applications via REST API, Boto3, or AWS CLI.

Enhanced security & data privacy

Scanned for security vulnerabilities, pre-trained models in AWS Marketplace help you gain control over your data and enhance security.

▶ Models are always deployed in network isolation mode with an option to deploy them in your Amazon Virtual Private Cloud (VPC) to implement additional security features that restrict and monitor traffic.

Flexible evaluation & pricing options

▶ Evaluate solutions at your own pace with try-before-you-buy options and interactive demo features.

▶ Pricing options like pay-as-you-go and annual contracts via Private Offers are available, giving you the flexibility you need regardless of the duration of use.

Close

Back to top »

How to deploy pre-trained models in Amazon SageMaker

Pre-trained models in AWS Marketplace can be deployed directly on Amazon SageMaker through a Jupyter Notebook, SageMaker SDK, or AWS CLI. This experience is powered by the AWS Marketplace catalog and APIs so you get all of the same benefits such as seamless deployment and simplified billing.

How to deploy pre-trained models in Amazon SageMaker
Close

Back to top »

  • Social distancing detector
  • Image classification model that analyzes social distancing in public areas

    Provectus’ Social Distancing Detector analyzes the distances between people based on their approximate height. This solution fits well into video analytic workloads for businesses that are looking to minimize disease transmission risks, gather data for decision-making processes, and ensure adequate personal space for their employees and clients.  

    Image classification model that detects social distancing violations in public areas’
  • Mask detector for epidemiological safety
  • Image recognition model to detect the absence of masks or respirators

    ViTech Lab’s Mask Detector for Epidemiological Safety improves safety in laboratories, healthcare facilities, educational institutions, industrial companies, military facilities and more. It is trained on a manually selected dataset and works with live footage from CCTV cameras. ViTech Lab’s Mask Detector for Epidemiological Safety detects mask and automatically notifies safety engineers of non-compliance.

    SOCIAL DISTANCING DETECTOR
  • Hard hat detector
  • Real-time detection of PPE compliance for factory floor settings and construction sites

    Prevectus’ Hard Hat Detector for Industrial Worker Safety is a computer vision-driven ML model designed to detect PPE compliance on a factory floor or construction site. This solution analyzes images to determine if a hard hat is being worn in real-time. Trained on data from CCTV cameras, this ML model can be used in oil & gas, manufacturing, construction, and steelmaking industries to help ensure safety and compliance.

    HARD HAT DETECTOR

Object detection

State of the art open source computer vision models trained using ResNet50, YOLOv3, Faster-RCNN to detect thousands of common category objects.

Detect multiple objects on the input image including category names, confidence scores, and absolute locations.

Image feature extraction and ImageNet category prediction model for highly efficient image classification.

A powerful network for fast and accurate object detection, trained on COCO dataset with 80 common object categories.

Close

Back to top »

Purpose built CV models

Pre-trained computer vision models that help automate data extraction and gather insights.

Automatically detects a car license plate and identifies the license plate characters.

Extracts key entities from a passport page to automate form filling.

Vehicle detector model for traffic analytics and intelligent transportation systems that operates from aerial and elevated perspectives.

Close

Back to top »

Text/Natural Language Processing models

Pre-trained models for text/natural language processing tasks like named entity recognition (NER) and text summarization.

Mphasis DeepInsights Text Summarizer

Text summarizer with the ability to reduce the size of long documents in to a few sentences.

Transaction Data Parsing (NER)

Extracts entities such as brand, color, category from supplied text.

Transaction Data Parsing (NER)

Detects 18 different entity types including buildings, works of art, landmarks, locations, and vehicles.

Close

Back to top »

How pre-trained models help solve real-world use cases

The AWS Marketplace hackathon took place in March 2020 and had thousand participants collaborating on 57 projects. Here are three stand out projects from the event that demonstrate how pre-trained models can be used to help solve real-world use cases.

A traffic monitoring and visualization platform that utilizes ML models to analyze mobility patterns.

A chat bot that utilizes ML models to support inferences and helps you at work, saving you time and boosting your productivity.

Camera-enabled IoT device powered by ML Models to detect if people are wearing masks when accessing restricted areas.

Close

Back to top »