Applications

Computer Vision

Detection, inspection, tracking and scene understanding

Computer vision is the most widely deployed edge AI workload, spanning quality inspection, safety monitoring, people and vehicle analytics, and robotic guidance.

Solutions in this category

Representative platforms with the details engineering and procurement teams shortlist on.

Object detection boxes drawn over a busy industrial scene
Ultralytics

Ultralytics YOLO11

Open-source vision model family covering detection, segmentation, classification, pose and oriented boxes.

Released
10 September 2024
Tasks
Detection, segmentation, classification, pose, oriented bounding boxes
Licence
AGPL-3.0, or Ultralytics Enterprise commercial licence
Deployment
Any edge device or server able to run the exported model
Industrial inspection
People and vehicle analytics
Agricultural monitoring
Developer screen showing multiple analysed video streams
NVIDIA

NVIDIA DeepStream SDK

GPU-accelerated SDK for building real-time video analytics pipelines on Jetson and x86 GPUs.

Type
Streaming vision AI pipeline SDK
Targets
NVIDIA discrete GPUs on x86 and NVIDIA Jetson platforms
Deployment
On premises, at the edge and in the cloud
Deployment
Jetson device, edge server or cloud GPU instance
Multi-camera video analytics
Traffic and city monitoring
Retail and safety analytics
Developer workspace with camera feed and code side by side
Roboflow

Roboflow Inference

Open-source inference package and server that turns a computer or edge device into a vision deployment target.

Type
Inference library and HTTP inference server
Licence
Apache 2.0 core library; separate licences for model and enterprise components
Features
Model loading, pre- and post-processing, CPU and GPU optimisation, workflows
Deployment
Workstation, edge device or on-premise server
Camera-based inspection
Prototype to production vision apps
On-premise inference services

Specifications are summarised from publicly published manufacturer material and are provided for orientation only. Always confirm current figures directly with the manufacturer before purchasing or designing in.

Companies working in this category

NVIDIA

Chip manufacturer

Develops the Jetson family of edge system-on-modules, the DGX Spark desktop AI system and the JetPack, TensorRT and DeepStream software stacks used for edge inference, robotics and vision AI.

Santa Clara, California, United States

Texas Instruments

Chip manufacturer

Supplies the AM6xA and TDA4 processor families, which combine Arm application cores, vision and imaging accelerators and a deep learning accelerator for embedded vision, robotics and automotive designs.

Dallas, Texas, United States

Hailo

Edge AI startup

Fabless company developing discrete edge AI processors, M.2 and PCIe accelerator modules and the Hailo AI Software Suite, spanning classic vision workloads and, with Hailo-10H, generative models.

Tel Aviv, Israel

Sony Semiconductor Solutions

Chip manufacturer

Produces the IMX500 intelligent vision sensor, a stacked CMOS image sensor with an integrated AI processing block and on-chip memory, supported by the AITRIOS edge AI sensing platform.

Atsugi, Kanagawa, Japan

Advantech

Hardware manufacturer

Industrial computing manufacturer supplying modular fanless edge systems such as the MIC-770 series and GPU-dense edge AI servers such as the AIR-520 for factory and infrastructure deployments.

Taipei, Taiwan

ADLINK Technology

Hardware manufacturer

Builds embedded computing modules and the DLAP family of deep learning acceleration platforms, including fanless Jetson-based inference systems with Power over Ethernet camera inputs.

Taoyuan, Taiwan

Aetina

Hardware manufacturer

Designs fanless edge AI systems and carrier boards around NVIDIA Jetson modules, including wide-input, wide-temperature platforms for machine vision and mobile robotics integrators.

New Taipei City, Taiwan

OnLogic

Hardware manufacturer

Builds configurable industrial computers, including fanless systems with Intel Core Ultra processors and expansion for Hailo accelerators or NVIDIA graphics cards.

South Burlington, Vermont, United States

Axis Communications

Computer vision company

Network video manufacturer producing cameras that run deep-learning analytics on the device through AXIS Object Analytics and the open AXIS Camera Application Platform.

Lund, Sweden

Hanwha Vision

Computer vision company

Video surveillance manufacturer whose cameras include an on-board NPU for object detection, classification and licence plate recognition, with AI-assisted noise and bandwidth reduction.

Seongnam, South Korea

Ultralytics

Software provider

Maintains the YOLO family of open-source vision models covering detection, segmentation, classification, pose estimation and oriented bounding boxes, dual-licensed under AGPL-3.0 and a commercial licence.

Frederick, Maryland, United States

Roboflow

Software provider

Provides computer vision tooling including the open-source Inference package and inference server, covering model loading, pre- and post-processing and workflow execution on edge devices.

Des Moines, Iowa, United States

Boston Dynamics

Robotics company

Robotics manufacturer building mobile and industrial robots that rely on on-board perception and autonomy rather than remote compute.

Waltham, Massachusetts, United States

Computer Vision: common questions

What limits real-time video inference at the edge?
Typically decode capacity, memory bandwidth and model complexity rather than raw accelerator throughput. Check the number of concurrent streams a platform is documented to sustain at your target resolution and frame rate.