Using Object Detection on IoT Devices

Something big is happening. The real world is being radically transformed by tech that has never had as much intelligence and capability as it does today. Sea changes are on the horizon and forward-thinking businesses are shifting their development focus to arguably the most significant change since the invention of the printed circuit board.

Artificial intelligence is here; consumers use it every day to help shop, get directions and search on the internet. Devices and products are getting smarter and more connected across industries including consumer electronics, manufacturing, healthcare, transportation, and more.

AI will eventually affect many more areas of daily life and one key segment of that radical shift is being powered by computer vision (CV).

CV is the part of AI that enables computers to analyze visual data and make decisions based on what they “see.” New technologies such as the Internet of Things (IoT) and 5G are in the process of converging and creating some fantastic new uses for that data. Whether you’re working with a connected IoT device or an unconnected embedded device on the edge, the ability to build and train machine learning AI in your app is more accessible than ever before.

One of the basic building blocks of CV is object detection, and it's an excellent place to start exploring some of the fascinating things that AI is poised to bring to everyday life.

Life on the High-Speed Edge

Object detection serves as the base for the foundation of some awesome capabilities. It decides if an object is present in an image or not, and if it is, determines where in the frame it’s located. The algorithm used to accomplish this task is relatively simple, but it produces powerful results.

In the IoT schema, sensors collect data continuously. It is more efficient to process this constant stream of data (i.e., from a video camera) on the sensors themselves rather than transport data across the network or onto the cloud. Some developers refer to these embedded devices as "the edge." Models and platforms are getting more efficient, allowing for more accurate onboard processing on low-power, resource-constrained edge devices.

Smart sensors such as cameras and other devices perform object detection and closely related tasks like object counting and classification. Doing this preprocessing work on the device increases efficiency by keeping unnecessary traffic off the network.

Examples of Applications Powered by Object Detection

The AI space is exploding as developers discover the potential of the tools now available. Here are a few of the more compelling uses of object detection and other computer vision services from a variety of different real-world use cases:

• Healthcare: Since most medical data is visual, object detection is a natural fit. Using computer vision in the healthcare field for analyzing X-rays and other forms of imaging is already helping healthcare professionals provide speedier and more appropriate treatment plans. Now, scope procedures such as endoscopies and other image-heavy operations allow the cameras being used to identify and determine certain classes of malicious or malignant cells instead of relying solely on the physician’s human eye.
  
  
• Security: Protection against crime and terrorism are at the forefront of object detection. Spotting out-of-place packages in airport terminals or other high-risk areas, tracking suspects without putting law enforcement in harm's way and alerting security to unauthorized entries are just the beginning..
  
  
• Agriculture: Farmers use object detection to identify invasive weeds and pests in their crops. An added health benefit is the ability to target only the pests that are present in the field, as they can reduce (and perhaps eliminate) the use of herbicides and pesticides.
  
  
• Manufacturing: Object detection plays a significant role in improving manufacturing tasks such as counting, sorting and quality control. AI contributes to productivity by helping manufacturers with predictive maintenance and plant automation.
  
  
• Industrial: Operators of remote storage facilities for petroleum and chemicals benefit from object detection systems that report leaks, spills or other problems in real-time. Workers are not tied up doing routine checks and can deploy to the site only when needed.
  
  
• Retail: Object detection provides retail store management with vital information about customer behavior, traffic flow in the store, and the popularity of displays and promotions. As retailers fully adopt the power of AI, they will be able to serve consumers with a vastly improved shopping experience.
  
  
• Smart Cities: Cities in the future will incorporate all forms of AI to provide a better living environment for their citizens. Object detection will play a critical role by providing a means to regulate traffic flow, provide protection against criminal activity, and help improve sanitation by identifying problem areas.