Home

Short Name

Locate and count items in an image with object detection

Short Description

Use deep learning to create a model and a REST endpoint to allow your app to detect, locate and count your product on store shelves

Offering Type

Cognitive & Data Analytics

Introduction

PowerAI Vision uses deep learning to create trained models based on images that you upload and label. You don't need to write any code to train, deploy, and test a new object detection model. You will simply upload the images, use your mouse to label the objects in your images, and then let PowerAI Vision do the learning. Once the model is built, you can deploy it with a single click. The deployed model can be tested in the PowerAI Vision UI. You can also use the REST endpoint with a curl command, your favorite REST client, or your own app.

Author

By Mark Sturdevant

Code

• https://github.com/IBM/powerai-vision-object-detection

Demo

• n/a

Video

• https://www.youtube.com/watch?v=xoLcXQs4SP4

Overview

In this Code Pattern, we will use PowerAI Vision Object Detection to detect and label objects, within an image, based on customized training.

This example can easily be customized with your own datasets.

An example dataset has been provided with images of Coca-Cola bottles. Once we train and deploy a model, we'll have a REST endpoint that allows us locate and count Coke bottles in an image.

Deep learning training will be used to create a model for object detection. With PowerAI Vision, deep learning training is as easy as a few clicks of a mouse. Once the task has completed, the model can be deployed with another click.

PowerAI Vision presents REST APIs for inference operations. Object detection with your custom model can be used from any REST client and can also be tested in the PowerAI Vision UI. An example node.js app is included to demonstrate how to upload an image and draw the image with labels and bounding boxes around detected objects.

When the reader has completed this Code Pattern, they will understand how to:

• Create a dataset for object detection with PowerAI Vision
• Train and deploy a model based on the dataset
• Test the model via REST calls

Flow

1. User uploads images to create a PowerAI Vision dataset
2. User labels objects in the image dataset prior to training
3. The model is trained, deployed and tested in PowerAI Vision
4. User can detect objects in images using a REST client

Included components

• IBM Power Systems: A server built with open technologies and designed for mission-critical applications.
• IBM PowerAI: A software platform that makes deep learning, machine learning, and AI more accessible and better performing.
• IBM PowerAI Vision Technology Preview: A complete ecosystem for labeling datasets, training, and deploying deep learning models for computer vision.

Featured technologies

• Artificial Intelligence: Artificial intelligence can be applied to disparate solution spaces to deliver disruptive technologies.
• Node.js: An open-source JavaScript run-time environment for executing server-side JavaScript code.

Blog

• https://developer.ibm.com/code/?p=29026&preview=true

Links

• Object Detection: Object detection on Wikipedia
• PowerAI Vision: Deep Learning and PowerAI Development
• TensorFlow Object Detection: Supercharge your Computer Vision models with the TensorFlow Object Detection API
• AI Article: Can Artificial Intelligence Identify Pictures Better than Humans?
• From the developers: IBM PowerAI Vision speeds transfer learning with greater accuracy -- a real world example