By 2027, ABI Research forecasts the installed base of Artificial Intelligence (AI)-enabled cameras to be 976.4 million, signaling a growing need for machine vision technologies. Today’s companies offering machine vision solutions are vital to the betterment of a number of applications, encompassing manufacturing, smart cities, building development, public safety, logistics, and transportation. An abundance of modern Machine Learning (ML)-based software products are embedded with edge ML capabilities that vastly improve a wide range of use cases, notably image-based inspection. This means a robotic system with an installed camera stacks pallets correctly or a flaw on a product's surface can be detected. To give you a better idea of how these systems work and who is providing them, this post highlights eight machine vision software companies that are helping enterprise customers work more efficiently.
Table of Contents
Why Machine Vision Companies Are Needed
As PowerReviews pointed out in a recent survey, 65% of consumers return products due to poor quality. This is why machine vision software solutions are a prime focus for manufacturers and logistics providers. Automated visual inspection is incredibly important for manufacturing and logistics players, as well as some other industries because the technology ensures high-quality products are being shipped and workflows are optimized. As noted in a recent Landing AI whitepaper, AI-powered quality inspection can detect anomalies up to 90% faster than manual inspection.
For manufacturers, a product recall would be a nightmare that concludes with a crippling brand reputation and plummeting sales. As labor shortages are already echoing throughout the industry, poor product quality is the last thing you need. And for logistics providers, it’s crucial that shipments don’t get damaged and that automated workflows are being carried out optimally. Ultimately, adopting machine vision software means that assembly line workers can increase their capacity while maintaining Quality Control (QC) and engineering teams can spend more time on the most pressing matters.
With that, let’s dive into the first machine vision company on the list.
Company #1: Arcturus
Toronto-based machine vision company Arcturus focuses on software, hardware, and solutions for machine vision (camera), voice, and secure Internet of Things (IoT). The three primary markets that the volumetric video capture startup targets are smart buildings and cities, intelligent transportation systems, and smart healthcare. Recently, Arcturus partnered with Kinara to leverage the latter’s Ara-1 Edge AI processor. The processor will be combined with Arcturus Brinq edge AI and vision analytics software to enable smart city applications. As an example, the press release mentions how a city bus embedded with smart cameras could identify potholes or anything else that interferes with a smooth ride—and then report the issues to city operation teams.
Arcturus’ product portfolio reflects its commitment to edge deployment in machine vision, such as the uCMK64-VoIP (Voice over Internet Protocol) module. This product conjoins the real-time performance of a microcontroller with the SIPxtream voice communications and the Mbarx Secure IoT. And because of its integrated signaling stack, multicast audio streaming, recorded announcement playback, and audio clarity, the uCMK64-VoIP module is a good match for intercoms, handsets, Public Address (PA) systems, or the conflux of radio-over-Intellectual Property (IP).
Figure 1: SIPxtream Voice and Video Services
The uCMK64-VoIP module is just one product from Arcturus that users can leverage. The company has machine vision software and hardware solutions that address the following categories:
- Edge Communications: Voice, video communications, signaling, codecs, and audio intelligibility.
- Edge Artificial Intelligence (AI): Object detection, image processing, classification, algorithms, model training, AI pipeline tooling, and workflows for public safety applications.
- Edge Connected Things: Secure management of connected devices, firmware, configuration, provisioning, and automatic updates.
- Cloud Edge: Cloud-native design and integration, containerization, Platform-as-a-Service (PaaS) software and microservices, frameworks, and integration platforms.
Company #2: Elementary
Elementary is a robotics startup founded by experts in intelligent/connected hardware in the IoT and robotics in mid-2017 under the name Elementary Robotics. In late 2022, Elementary closed US$30 million in Series B funding so the team could meet customer demand and scale across more regions. The company, headquartered in Los Angeles, has developed a software platform that enables the logistics and manufacturing industries to apply Machine Learning (ML)-enabled machine vision to day-to-day operations. Elementary offers easy-to-use drag-and-drop software that fuses AI tools with traditional vision tools for inspection via camera analysis.
Elementary's no-code camera setup means that cameras can be configured from anywhere, including on-site and remotely, with no coding experience required. Plus, the settings can be saved to the cloud and used in other inspections and locations. Moreover, training the ML model is only a matter of feeding the system a few examples of what a human eye would look out for when inspecting. Best of all, because Elementary’s machine vision software is entirely data-driven, the anomaly detection tools can seamlessly migrate across various use cases and different industries.
Figure 2: ML-Based Machine Vision Software
Company #3: Instrumental
The software that Instrumental, an AI-based machine vision company founded by two former Apple engineers, offers is aimed at automating object detection and engineering workflows. More specifically, manufacturing assembly lines are a key focus of the startup’s motivation. The customers of Instrumental, which could stand to save billions of dollars in scrap and rework costs, use the ML software to identify and solve issues on their assembly lines. In February 2022, Instrumental’s Research & Development (R&D) was supplemented by US$50 million in Series C funding. Some of the top investors are BAM Elevate, Canaan, Root Ventures, and Eclipse Ventures.
The ML processes are able to assess every unit on a line and pinpoint patterns in images, measurements, and test data—then identify irregularities or other quality problems; doing so would be impossible with only human workers. Using Instrumental’s self-service Application Programming Interfaces (APIs), manufacturers can collect source data, images, and videos from any source, so that they can understand where inefficiencies and failures exist. This enables customers to get out ahead of a problem before it leads to significant downtime or financial loss. Recent research suggests that human error is a more common cause of downtime in manufacturing than in other sectors and that this unplanned downtime costs companies roughly US$50 billion every year.
Figure 3: Comparing Human Vision with Computer Vision
In one notable case study from the startup's website, Instrumental’s AI-powered machine vision software platform was leveraged by technology and weapons manufacturer Axon. After just one day of deployment, Axon already discovered more than 20 different kinds of issues that were previously unknown. It also didn’t take long to reduce defect rates by 6% as the design and processes were greatly improved. One final note is that the engineering team saved 35 hours of development because it began focusing on more important tasks.
Figure 4: Axon Fleet 3
For more on the latest technologies being leveraged in the manufacturing industry, read our whitepaper, 74 Technology Trends To Expect in 2023.
Company #4: Mech-Minded Robotics
Mech-Minded Robotics is a Chinese-based company that has extensive experience in computer vision, ML, software engineering, and robotics. As can be inferred, these machine vision solutions are highly beneficial in warehouse settings and factory settings. Considering that just 6% of warehouses and Distribution Centers (DCs) are currently using Automated Guided Vehicles (AGVs)/robotics, vision inspection platforms like the one offered by Mech-Minded Robotics will continue to grow in demand as adoption widens. And labor shortages could certainly serve as a catalyst; however, economic uncertainty may hinder robotics implementation plans in the next couple of years.
The company’s Three-Dimensional (3D) vision-guided robotics solutions for logistics and manufacturing can perform the following tasks:
- Depalletizing
- Palletizing
- Bin picking
- Sorting
- Assembling
- Locating
- Gluing parcel picking
Figure 5: Vision-Guided Robot Grabbing Workpieces on an Automotive Assembly Line
(Source: Mech-Minded Robotics)
All are based on using 3D sensors for robotic deployment, but the most notable machine vision solutions on offer from Mech-Mind Robotics are Mech-Eye Industrial 3D Camera, Mech-Vision Vision Algorithm Software, and Mech-Viz Robot Programming Environment. These three products are part of the firm’s AI + 3D Vision + Industrial Robot Integrated Solutions lineup.
Additionally, the Mech-DLK Deep Learning (DL) software is also an important solution that enables warehouses and factories to bolster productivity and polish product quality. Integral to making this solution is the combination of tailored ML algorithms and a Graphical User Interface (GUI) that allows users to quickly train models and run demanding applications.
Company #5: Landing AI
Landing AI, a company founded by Dr. Andrew Ng, which made Forbes’ list of America’s Best Startup Employers 2022, offers an end-to-end AI software platform called LandingLens. This machine vision platform has industrial customers in mind, empowering them to quickly build, deploy, and scale ML-powered visual inspection solutions using data-centricity. Here, the emphasis is on data, not code.
According to Landing AI, enterprises from various industries, such as automotive, electronics, and medical device manufacturing, have experienced better results with AI- and DL-based computer vision solutions when compared to traditional, rules-based implementations. If an algorithm is working with parts that have convoluted features or obscure part defects, rules-based machine vision may run into trouble. For example, in working environments where new or custom parts are introduced or lighting conditions fluctuate, rules-based algorithms could produce a high rejection rate because the technology has trouble differentiating between genuine defective parts and acceptable variation. But AI- and ML-based solutions are more capable of an accurate inspection in these scenarios.
Figure 6: “Is This Really a Scratch?”
All in all, the LandingLens software platform enhances visual inspection and reduces the frequency of false positives—leading to improved product quality. Furthermore, the machine vision startup reduces development time and is a conduit for quickly scaling projects as it also standardizes the development of DL solutions. This latter point translates to less debate on defect labeling in machine vision systems, which is inputted by human workers and can often lead to disagreement. When teams use a consistent means for aggregating and labeling images, and for training, optimizing, and updating the models, collaboration is far more effective.
For the last three sections, we’ll explore machine vision software from the major chipset companies Intel, NVIDIA, and Qualcomm.
Company #6: Intel
The OpenVINO toolkit and Geti platform are software offerings from Intel geared to enterprise machine vision users. The company’s Geti platform, announced in September 2022, makes life a lot easier as it relates to labor-intensive data upload, labeling, training, model optimization, and retraining tasks. As the platform is compatible with OpenVINO, MV models can be fine-tuned for machine vision at the edge.
Company #7: NVIDIA
NVIDIA, as it expands far beyond its gaming roots, offers three software platforms for edge ML machine vision: Clara, Isaac, and Metropolis. Clara, a scalable, software-defined platform, is used in the healthcare sector for real-time visual data processing at the clinical edge. Isaac is the company’s data-driven robotics simulation application that accurately represents the physical world in a virtual medium. Because of its ability to develop, test, and manage AI-based robots, Isaac is well-suited to improve the vision systems of mobile robots—key to workplace safety. Finally, there’s Metropolis, with the mission to fill the void between visual data and AI so that operational efficiency is improved and industrial workplaces keep employees safe.
Company #8: Qualcomm
Qualcomm’s AI Engine Direct is an AI library that deploys existing models to the AI accelerators on Qualcomm Technologies’ platforms. Developers will find it useful to develop one single feature, then replicate the model across various products and tiers. AI Engine Direct falls under the umbrella of Qualcomm AI Stack Portfolio, which is the broader software lineup customized for Qualcomm Technologies-powered devices.
How ABI Research Can Help
As part of the company’s AI & Machine Learning Research Service, ABI Research analysts are in the vanguard of innovation in the machine vision space. Our thought leadership content consistently finds its way into top-tier media outlets as our analysts accurately portray the technology markets and provide strategic recommendations on the next steps—whether that’s for implementation or developing/marketing these products.
To learn more about machine vision software and services—and the companies/startups providing these products—download the report. The report goes into a little more detail on these listed companies, as well as recent developments from companies like Veo Robotics, Invisible AI, Wahtari, Pensa Systems, AiFi, Netradyne, Atos, Standard AI, and more. Or if you're not ready for the report yet, read our Edge Machine Learning for Computer Vision in the Age of Industry 4.0 Research Highlight.
