Computer vision
Image and video processing, object recognition and quality inspection.
Discuss your task
Possible applications
Object recognition, video event detection and visual quality inspection are possible applications.
Scope of work
Stream assessment, labelling, model selection and training, event integration into workflows.
Data and prerequisites
Representative images and video, processing rights and object definitions. Lighting and camera positions affect quality.
Integration and deployment
Cameras, video servers and operational systems. Processing at the edge or on central servers.
How we evaluate results
Precision, recall, false positives, missed events and latency on a separate dataset reflecting operating conditions.
Before we start
Will any camera work?
Quality depends on resolution, angle, frame rate and lighting. We test feasibility on samples from your streams.
Do we need labelled data?
Verified examples of objects and events are usually needed. The volume depends on the task and approach.
What would you like to solve?
Describe a process or problem. We will discuss an approach, the data needed and the next step.
A formal specification is not required. Tell us what happens today and what you would like to improve.
admin@invise.org