AI Inspectors

Configured AI models deployed on specific inspection workflows or stations in LIVIS. Each AI Inspector includes the model, inspection logic, threshold settings, and deployment parameters, acting like a digital quality control agent.

AI-Assisted Labeling

Semi-automated labeling feature where a pre-trained model suggests annotations to accelerate the annotation process, especially in repetitive or high-volume tasks.

Annotation History

A version-controlled log of all annotation changes, including who made changes and when, enabling audit trails and rollback capabilities.

Annotation Insights

Visual and statistical summaries of annotation trends, label consistency, and annotator performance to improve labeling accuracy and efficiency.

Appliance

A dedicated on-premise hardware system configured to host the LIVIS Platform for AI model development, training, and testing. It enables customers to build and manage vision models securely within their local infrastructure without relying on cloud access.

Centralized Workspace

A shared cloud environment within LIVIS, enabling multiple teams or users to collaborate, manage, and analyze datasets and AI models centrally, promoting consistency and resource optimization across the organization.

Controller

A specialized edge device that hosts and executes LIVIS Edge+ software. It manages real-time processing, AI model deployment, and interactions with cameras and other industrial equipment on-site.

Data Backup

A scheduled or event-driven process in LIVIS that securely backs up all relevant data—including models, training sets, annotations, deployment configs, and inference logs—specific to a customer’s on-premises (hub) deployment. This ensures disaster recovery, rollback, and migration support without data loss.

Dataset Analytics

Dashboard providing metrics like class distribution, annotation coverage, and data quality to assess dataset readiness and identify imbalances or labeling gaps.

Dataset Management

Centralized interface in LIVIS to upload, organize, version, and manage image and video datasets for training and validating AI models. Supports tagging, filtering, and access controls.

Dedicated Workspace

An exclusive cloud environment allocated to a single customer or user group within LIVIS, ensuring customized resource allocation, enhanced security, and dedicated performance for specialized needs.

Deployment Recipe Manager

A tool in LIVIS to configure and manage deployment settings such as model version, thresholds, preprocessing steps, and output format for each inspection station.

Device Management

Interface to manage edge devices, cameras, and inspection stations. Includes configuration, status monitoring, software updates, and model distribution controls.

Fine-Tuned Models

Custom AI models trained or adapted from foundation models using user-specific datasets to detect particular defects or patterns unique to a manufacturing line.

Foundation Models

Pre-trained purpose-driven models available in LIVIS that can be fine-tuned on specific manufacturing datasets, offering a fast start for new inspection use cases.

GPU Hours

The unit of compute resource usage in LIVIS, representing one hour of GPU time consumed for training, inference, or evaluation tasks. GPU hours are tracked per project or organization for usage reporting and billing.

Image Augmentation

Built-in tools to generate synthetic variations of training images (e.g., rotation, flip, brightness changes) to enhance model generalization and reduce overfitting.

Isolated Workspace

A fully segregated cloud environment within LIVIS designed for highly secure and private operations, ensuring maximum protection, confidentiality, and dedicated resource utilization, typically necessary for sensitive or proprietary workflows.

LIVIS Evaluate

A module in LIVIS designed for benchmarking and validating AI model performance. It provides confusion matrices, precision/recall curves, defect-wise accuracy, and comparisons between multiple models on a given dataset.

Model Monitoring

Real-time monitoring of deployed models to track metrics like inference time, accuracy, and drift detection to ensure continued model performance in production.

Report Assets

Templates, visual dashboards, and data exports generated from inspection results in LIVIS. These assets include defect summaries, trend analysis, pass/fail rates, and compliance documentation tailored for different stakeholders.

Role Based Access Control (RBAC)

A security mechanism in LIVIS that restricts platform access based on user roles (e.g., Admin, Inspector, Reviewer). Each role has predefined permissions to ensure data integrity and operational security.

SSO (Single Sign-On)

A user authentication method in LIVIS that allows users to log in using corporate credentials (e.g., Google Workspace, Azure AD), enabling centralized access management and improved security.

Training Assets

All digital resources used for training AI models in LIVIS, including labeled images, annotation schemas, preprocessing settings, and augmentation configurations. These assets form the foundation of the model training pipeline.

Video Annotations

Feature allowing users to annotate objects or defects across video frames. Useful for tracking defects over time or capturing frame-by-frame changes in motion-based inspections.