EVA SDK | Edge AI Vision Analytics Development Kit | ADLINKEVA SDK | Edge AI Vision Analytics Development Kit | ADLINK

Fast and Intuitive

EVA SDK is edge computing visual analysis software supporting single source AI machine vision project deployment across different hardware platforms. EVA SDK is embedded in ADLINK leading edge AI devices to accelerate edge AI application development in smart manufacturing.

Supports more than ten camera protocols and field-ready application plugins

Supports TensorRT^™, OpenVINO^™ and ONNX Runtime AI inference frameworks

Low-code/no-code GUI simplifies edge Al vision PoC for fast AI inference validation

EVA SDK Features

EVA Runtime & EVA IDE (Integrated Development Environment) included
Intuitive GUI for fast and easy AI inference pipeline development
10+ supported camera protocols reduce integration effort
5+ vertical oriented sample code and templates provide faster application development (Geofence, Wear Detection, Cookie Inspection, Parts Preparation, Parts Assembly, etc.)
Rich app-ready samples and support available with exclusive EVA Portal
Over-the-air model update powered by CarOTA
Docker container support

Find EVA supported hardware Get more EVA SDK technical resources

Targeted Applications

Hazmat offload safety management

Semiconductor factories commonly have restricted areas that require workers to wear protective safety gear when offloading hazardous cargo. In this use case you can see how a NEON smart camera with EVA SDK can monitor personnel for appropriate safety gear and immediately sound an alarm to prevent accidents if personnel enter dangerous areas without protective equipment.

Download Use case

Robot Hazard Prevention

Factories commonly have restricted areas to keep workers safe from moving automation or robots. In this use case you will see how a robot working area is designated as a restricted area to prevent people from entering it. If someone enters the restricted area, the region will turn red if the display is set to true, and an alert event will be recorded as metadata for a downstream element to activate.

Download Use case

Assembly SOP Compliance

EVA is used for monitoring product assembly based on a standard operating procedure. Each part requires assembly according to a predetermined time with the total assembly time for multiple parts analyzed to dynamically adjust part assembly in real-time. AI-based vision controls monitoring and guides operators to follow the adjusted SOP both through real-time on-site reminders and long-term data analysis.

Download Use case

Fab Clean SOP Compliance

High-precision semiconductor factories use air showers to remove surface-deposited particles from personnel prior to them entering a cleanroom, but workers failing to follow proper cleanroom procedures can put products at risk. Using AI to detect whether an employee's behavior complies with SOP can reduce the amount of dust by 80% and improve production quality.

Download Use case

Personal Protective Equipment Detection

In order to protect personnel in the field who need to access and operate high-voltage equipment, a foolproof safety system is necessary. In this use case, you can learn how EVA uses AI vision to immediately detect whether an operator is wearing the correct safety equipment, avoiding personal injury.

Download Use case

ADLINK provides reference codes for you to easily deploy these applications.

Leverage the Benefits of an All-in-One Edge AI Vision Platform

Extensive image Acquisition capability reduces integration effort

Optimized AI Inference models reduce development TCO

Multiple image pre-processing plug-ins enhance image quality quickly and easily

High Connectivity from Edge to Cloud with rich output.

Supports Hybrid AI Inference Engines Simultaneously in a Pipeline

Real-time pipeline execution result visualization makes editing and debugging easier.

EVA supported AI models:

Network Original framework AI model type GoogLeNet Caffe/ONNX Classification ResNet18 Nvidia TLT ResNet ONNX SqueezeNet VGG MobilenetV2 AlexNet CaffeNet RCNN_ILSVRC13 DenseNet-121 Inception_V1/V2 ZFNet-512 ShuffleNet_V1/V2

Network	Original framework	AI model type
GoogLeNet	Caffe/ONNX	Classification
ResNet18	Nvidia TLT
ResNet	ONNX
SqueezeNet
VGG
MobilenetV2
AlexNet
CaffeNet
RCNN_ILSVRC13
DenseNet-121
Inception_V1/V2
ZFNet-512
ShuffleNet_V1/V2

Network	Original framework	AI model type
SSD Inception-V2	Tensorflow	Object Detection
Face detection(SSD based)	Caffe
Yolov3	Darknet/ONNX
Yolov4	Darknet/ONNX
Yolov3-tiny	Darknet
DetectNet_v2-ResNet18/34	Nvidia TLT
OpenPose	PyTorch	Pose Estimation
FCN-Alexnet	Caffe	Segmentation
Intel Segmentation	Caffe	Segmentation