Define requirements
Confirm inputs, hardware form, interfaces, power constraints and intended use.
AIBox event verification workflow
Select a step to explore the workflow. This walkthrough does not run inference.
Workflow step 01
CV screening
The CV module screens the input and produces candidate events for review.
Workflow step 02
Large-model verification
A large model reviews candidate events to assess whether they are valid.
Workflow step 03
Review results and feedback
Review results support the operational workflow, and relevant data feeds the training platform.
Model training workflow
From feedback samples to CV or VLM training.
Workflow step 01
Receive feedback data
The platform receives review data from AIBox to prepare samples for a target task.
Workflow step 02
Annotation and image generation
Automatic annotation and image generation provide candidate samples. Training data versions are published after human review.
Workflow step 03
Iterative training
Train or retrain a CV or VLM model, then use evaluation results to select a version.
Edge and smart-device delivery
For edge inference, interactive and specialized devices, with validation scoped to the current product stage.
Prototype and pilot validation
Use available prototypes or devices to evaluate models, interactions and workflows.
Define the product baseline
Use evaluation to define software versions, enclosure, thermal design and acceptance scope.
Delivery and operation
Scope delivery, updates, device management and maintenance according to product maturity.
Product enquiries
Tell us which product you need, your current equipment and the problem you want to solve.