Visual Inspection

deviceWISE AI Visual Inspection is a tool that can annotate images for defect detection, anomaly detection and quality control. It provides Visual Inspection capabilities to deviceWISE by utilizing advanced AI computer vision algorithms for Object Detection, Anomaly Detection and Text Recognition.

Introduction

Using advanced algorithms and being usable through an user-friendly interface, deviceWISE AI Visual Inspection allows to easily training and deploying machine learning models to identify defects and anomalies in real-time.

deviceWISE AI Visual Inspection is a tool that positions itself as a solution for engineers, technicians and quality control professionals, helping to streamline the inspection process and improve product quality as a whole.

What kind of inspection does Visual Inspection do?

  • Object Detection: Detects different objects within a set of images. This type of project deals with the detection of elements of a specific object to be defined within the training.

  • Anomaly Detection: For Anomaly Detection, the tool tries to find within a set of images, elements that are irregular within that set of images.

  • Text Recognition: For Text Recognition, the tool tries to identify a set of letters and words within the set of images provided. The Training Algorithm used in this case is a pre-trained Optical Character Recognition (OCR) Model.

Which are its components?

Studio - Used to create models and manage models;

Edge - Used with run inferences of given models.

System Requirements

Under this title you will find details about the System Requirements:

Hardware:

  Description
GPU

  • NVIDIA Tesla T4

  • NVIDIA Tesla A2

  • NVIDIA Tesla L2

  • Jetson XAVIER

  • Jetson ORIN
Processor A modern, multi-core processor is recommended for efficient processing of tasks.
Memory

A minimum of 8 GB RAM is recommended. More memory may be required for handling large datasets or complex models.
Storage

At least 10 GB of free storage space for the software installation, additional space for project data.

Operating Systems:

LINUX: deviceWISE AI Visual Inspection is supported with all versions of Ubuntu. This includes both LTS (Long Term Support) and standard releases. Make sure that the Ubuntu environment is running with proper functionality.

  • Ubuntu 18;

  • Ubuntu 20;

  • Ubuntu 22.

WINDOWS: deviceWISE AI Visual Inspection is also supported by Windows, Docker Desktop must be installed.

Click here for detailed instructions on Docker Desktop.

A dedicated Graphics Processing Unit (GPU) is recommended for optimal performance. NVIDIA GPUs are preferred due to their CUDA support. Make sure the machine has proper drivers for the GPU.

Jetson Device Compatibility

deviceWISE Visual Inspection software is compatible with NVIDIA Jetson devices. This includes models such as Jetson Nano, Jetson Xavier NX, and Jetson AGX Orin. Lenovo SE70 and Siemens Tensorbox 520 were also validated.

Users should have the latest drivers and software for their NVIDIA GPUs or Jetson devices. This includes CUDA Toolkit and cuDNN libraries GPU acceleration.

Ensure Docker is installed, version 20.0 or higher, for container management and deployment.

Install the latest NVIDIA Container Runtime to enable Docker containers to access NVIDIA GPUs, following the installation guide provided by NVIDIA.

What's Inside