Computer Vision
Image and video models in Python for detection, classification and inspection, scoped after we review sample images.
Custom quote after a short consultation · Timeline agreed in your quote
Overview
Computer vision lets software interpret images and video: counting items, spotting defects, classifying products or detecting objects in a scene. RF ETS develops these models in Python, drawing on an engineering background that includes cameras alongside sensors, embedded hardware and robotics.
Results depend heavily on the images, the quality of labels and the conditions in which pictures are taken, so this service is quote-only. We begin with a sample review. You send representative images or video, we check lighting, angles, resolution and labelling needs, and we then propose an approach, an evaluation method and a written quote.
A typical project covers image preparation, labelling guidance or label checks, model training and testing on images the model has not seen. You receive the trained model, inference code that runs on new images, an evaluation report showing where the model performs well and where it struggles, and notes on capturing better images in future.
Who it’s for
- Manufacturers that want to automate visual checks for defects or missing parts
- Robotics and automation projects that need object detection from a camera feed
- Retail and inventory teams that want to classify or count items from photos
- Research and product teams prototyping an image-based feature
Problems it solves
- Reduce repetitive manual inspection of images or video
- Find out whether your images are good enough before investing in a full build
- Get a working prototype that runs on new images
- Understand which conditions cause the model to struggle
What RF ETS delivers
- Sample review with recommendations on images and labels
- Prepared and checked training dataset
- Trained detection or classification model
- Python inference code for running the model on new images or video
- Evaluation report with performance on unseen images and failure examples
How we work
Sample review
You send representative images or video and describe what must be detected. We check quality, conditions and labelling effort.
Scope and quote
We propose the model type, evaluation method and deliverables in a written quote.
Prepare, label and train
We prepare images, check or guide labelling, then train and tune the model.
Evaluate and hand over
We test on unseen images, report results and failure cases, and hand over the model and inference code.
What we need from you
- Representative sample images or video from real conditions
- A clear definition of what counts as each class or defect
- Existing labels if available, or a person who can confirm labelling decisions
- Details of where the model will run, such as a PC, server or camera setup
How pricing works
This work varies too much for fixed packages. We scope it with you, then send a written quote with price, timeline and deliverables before any work starts.
Tell us what you need
Use the quote form or book a consultation and describe your goal, constraints and deadline.
We scope it with you
We review your material and agree deliverables, assumptions and what is out of scope.
Written quote
You receive a fixed price or milestone plan, timeline and revision terms before any work starts.
Not included
- Third-party costs: domains, hosting, paid plugins and themes, software licences, API or AI-model usage fees, data-provider credits
- Work outside the written scope agreed before work starts (handled as an add-on or a custom quote)
- Ongoing support after the delivery and launch-support window unless a support package is bought
Need something different?
Tell us what you need, and we’ll prepare a solution and pricing based on your requirements.
Computer Vision: common questions
How is a computer vision project priced?
This service is quote-only. After reviewing your sample images we quote based on image volume, labelling effort, model complexity and deliverables. The scope and price are confirmed in writing before build work starts.
Can you promise a detection accuracy?
No. Performance depends on image quality, lighting, angles and how consistent the labels are. We measure and report results on images the model has not seen and point out conditions where it is weaker.
Do you label the images for us?
Labelling can be included, or we can guide your team and check their labels. Which option suits you depends on volume and domain knowledge, and it is set out in your quote.
Can the model run on a device instead of a server?
Sometimes. It depends on model size and the hardware available. Tell us where it needs to run during the sample review and we will advise whether that is realistic and include it in the scope if so.
What if we need changes after delivery?
Revisions within the agreed scope are covered as stated in your quote. New classes, new camera setups or retraining on fresh data are priced as an add-on or an updated quote.
Often combined with
Machine Learning & Predictive Analytics
Custom machine learning and predictive analytics in Python, from data review and feature work to a tested model and clear documentation.
Custom quoteView service →Embedded & RoboticsRobotics & ROS Development
ROS and ROS2 development for prototype, simulation and research robotics: nodes, sensor integration, control and navigation logic, quoted per project.
Custom quoteView service →Embedded & RoboticsIoT Solutions
Prototype and pilot-scale IoT systems: ESP32 and Arduino devices, data transport, storage, dashboards and alerts, scoped through a custom quote.
Custom quoteView service →