Image classification
Categorising images against your defined classes with per-class accuracy reported.
Images and video
Computer Vision & Image AI
Detection, classification and inspection on images and video, deployed where the work happens.
What this is
Vision systems succeed or fail on data and deployment conditions, not model choice. We assess your real images, including the lighting, the angles and the edge cases, before scoping. Then we build detection, classification or verification pipelines that run in the cloud or on-site as the use case requires.
The problem
What this service addresses
If more than one of these is familiar, there is usually a case worth scoping.
- Visual checks are done manually and consistency varies by person and shift.
- Image volume has outgrown the team that reviews it.
- Defects or issues are caught late, after cost has already been incurred.
- Photographic evidence is collected but never systematically checked.
Scope
What we deliver
Every item below is a deliverable you receive, not an activity we perform.
- Feasibility pilot on your real images before full commitment
- Detection, classification or verification model fitted to your conditions
- Annotation strategy and labelled dataset with documented guidelines
- Inference pipeline deployed in cloud or at the edge as required
- Confidence thresholds with human review for uncertain cases
- Accuracy reporting by condition, covering lighting, angle and defect type
- Retraining process as new cases and edge cases appear
Capabilities
What the system does
Object detection and counting
Locating and counting items, with handling for occlusion and overlap.
Visual inspection and defect detection
Flagging deviations from an expected state, tuned to your tolerance for false alarms.
Document and identity image checks
Verifying that submitted images meet quality and content requirements.
Video analysis
Frame sampling and event detection over recorded or streaming footage.
Edge deployment
On-device or on-premise inference where bandwidth, latency or privacy requires it.
Applied
Where it is used
Each of these is a bounded job with a measurable result, which is how we prefer to scope a first build.
01
Quality inspection
Checks output against expected appearance and routes suspected defects for review.
02
Delivery and condition evidence
Verifies that proof-of-delivery or condition photographs are usable and show what is required.
03
Inventory and shelf monitoring
Counts stock and identifies gaps from fixed-camera or handheld images.
04
Document image quality
Rejects unreadable uploads at capture time instead of failing later in processing.
05
Site and safety monitoring
Detects defined conditions in footage and raises alerts with the frame as evidence.
Architecture
How it is put together
An indicative architecture. The final design follows your systems, data boundary and constraints.
-
01
Capture
Fixed cameras, mobile uploads, scanners or existing video infrastructure.
-
02
Preprocessing
Normalisation, quality gating, cropping and augmentation.
-
03
Model layer
Classification, detection or segmentation models selected for the deployment constraints.
-
04
Inference deployment
Cloud GPU batching, or edge devices where latency, bandwidth or privacy demands it.
-
05
Review workflow
Confidence-based routing into a human review queue with the image attached.
-
06
Feedback loop
Reviewed cases returned to the labelled set to improve subsequent versions.
Security & governance
Controls we build in
Access, privacy and oversight are part of the build rather than a later hardening phase.
- Images containing people handled under a defined lawful basis and retention limit.
- Accuracy reported by condition, so performance in poor lighting or unusual angles is visible.
- Human review mandatory where a visual decision affects a person or a commercial outcome.
- Edge deployment offered where images should not leave your premises.
- Annotation guidelines documented so labels remain consistent across contributors.
Delivery
Our approach
How we run the engagement
Discovery
Process walkthroughs, data review, constraints and a written scope you sign off before build.
Architecture
Data model, integration boundaries, hosting and the security model, agreed up front.
Proof of concept
A narrow build on your real data against an agreed quality bar, so the go-ahead is evidence-based.
Development
Two-week increments, each ending in a working deployment and a review.
Testing
Automated coverage on critical paths, evaluation sets for AI behaviour, acceptance testing with your users.
Deployment
Staged rollout into your infrastructure with monitoring and a tested rollback path.
Monitoring & support
Quality, cost and latency tracking after launch, with handover documentation or an ongoing retainer.
Proven work
Products we have built and sold
We are product builders as well as a services team. These are our own commercial platforms, not client references.
Digital sales & support agent
SIZH Virtual Agent
A virtual digital agent for sales, support and reception — trained on your business, live on WhatsApp, web, voice and email.
View product Mobility platformRideOn
A ride-hailing and fleet management platform, now with agentic dispatch, driver support and demand intelligence.
View product Multivendor marketplaceUniBooker
A configurable multivendor service-booking marketplace with agentic onboarding, matching and support.
View product
Questions
Computer Vision & Image AI — FAQs
The questions we are asked most often about this service.
How many images do we need?
Typically hundreds to low thousands per class for reliable results, and crucially they must cover your real variation. A thousand images in identical conditions is worth less than three hundred that span your actual range.
Do you always start with a pilot?
For vision, yes. Feasibility depends heavily on image quality and variation, and a short pilot on your real data is the only honest way to scope the build.
Can it run without internet access?
Yes, via edge deployment on suitable hardware. There are accuracy and model-size trade-offs, which we quantify during the pilot.
What accuracy should we expect?
We will not quote a figure before the pilot. Afterwards we report accuracy per class and per condition, and you decide whether it clears your bar.
Related
All AI First services
Services that pair with this
Intelligent Document Processing & OCR
Invoices, forms and contracts read, validated against your rules and posted into your systems.
Read moreMachine Learning Development
Predictive and classification models trained on your data, evaluated honestly and deployed as services.
Read moreMLOps & AI Deployment
Versioning, evaluation pipelines, deployment, rollback and monitoring for AI systems in production.
Read more