01 / OBJECTS & ENVIRONMENTS
Everyday visual context
Scope objects, activities, environments, commerce scenes, food, products, signage, or other real-world visual categories relevant to the model task.
Commissioned image data for AI training
NDPC REGISTERED · NDPC/DCP/13596
Registered with the Nigeria Data Protection Commission. Certificate available on request. See our Compliance page and Privacy Policy.
African image data for models that need real-world context.
BSG DataWorks scopes image-data projects around the objects, environments, documents, capture conditions, and quality criteria your computer-vision system must handle. We collect and prepare commissioned image data to the project brief, with the consent, provenance, annotation, and delivery questions made visible from the start.
Project scope
Commission image data around the visual task—not a generic image catalogue.
We begin with the model's visual problem and confirm the practical scope before collection or preparation starts. The project can define subjects, environments, variation, capture rules, annotation, quality checks, and delivery requirements.
02 / DOCUMENTS & TEXT IN SCENE
Visual documents and signage
Define document types, handwriting, printed text, signage, page conditions, image quality, and any text or metadata requirements for the project.
03 / CAPTURE VARIATION
Devices, angles, and lighting
Set the capture conditions, device mix, framing, lighting, backgrounds, viewpoints, and edge cases that make the image set relevant to the intended deployment context.
Computer-vision use cases
Image data designed around what the system needs to see and decide.
Tell us the visual task, the operating context, and the quality evidence your team needs. We use those facts to shape a collection or preparation brief rather than implying one dataset fits every computer-vision problem.
Object detection and classification
Define the objects, categories, scene conditions, occlusions, variations, and annotation rules needed for the target recognition task.
Document and visual-text intelligence
Scope documents, handwriting, labels, signage, layout conditions, language context, and the extraction or review task the system must support.
Retail, commerce, and operational scenes
Build visual examples around products, shelves, markets, food, workflows, and other contexts that need to reflect the agreed project setting.
Visual QA and evaluation
Prepare held-out examples, agreed label guidance, review criteria, and documentation to help the buyer evaluate the image-data scope against its own requirements.
Technical and quality requirements
Specify the image conditions that make a collection useful.
A strong image-data brief explains the visual task, what should and should not appear in frame, how variation is handled, how labels are defined, and what evidence accompanies delivery.
Capture plan
- Object, scene, document, or activity categories
- Locations, settings, backgrounds, and inclusion or exclusion rules
- Devices, framing, viewpoints, lighting, and resolution requirements
- Target variation, edge cases, and sampling considerations
Annotation and metadata
- Classification, detection, segmentation, text, or other label schema
- Written annotation guidance and issue-escalation process
- Metadata fields relevant to the buyer's model and evaluation needs
- File naming, folder structure, and delivery-format requirements
Quality assurance
- Image usability checks agreed for the project
- Label-review and acceptance-criteria process
- Sampling, feedback, correction, and version-control approach
- Clear treatment of out-of-scope or unusable material
Consent and provenance
- Source and collection-route documentation appropriate to the scope
- Written consent workflow when identifiable contributors are involved
- Early flagging of biometric, sensitive-setting, or re-identification questions
- Buyer-facing provenance and data-handling notes agreed for delivery
From visual brief to delivery
A practical route for commissioned image-data work.
We use the visual task and delivery context to make collection, preparation, annotation, quality, and governance decisions visible before production work begins.
Define
Clarify the model task, object or scene categories, intended operating context, exclusions, and the visual variation the project needs.
Design
Agree the capture rules, device and lighting conditions, label schema, metadata, quality thresholds, and evidence requirements.
Prepare
Collect or organise the scoped images, then annotate and review them using the agreed guidelines and quality process.
Deliver
Provide the image files, labels, metadata, and agreed documentation in the required structure and delivery route.
Frequently asked questions
Questions about African image data.
Do you offer a ready-made image-data catalogue?
No. Most image-data work is commissioned around the buyer's object classes, scenes, capture conditions, annotation requirements, and delivery needs. We confirm the practical scope before committing to a project.
Can a project include people in the images?
Potentially, but this is a special project question. The brief should explain why identifiable people are needed, how the images will be used, and what safeguards are required. We use written consent workflows where applicable and flag biometric, sensitive-setting, and re-identification questions early; buyers should obtain independent professional advice for their own legal and governance obligations.
Can you work with images we already have?
Yes. If you already hold source images, we can discuss a preparation, annotation, quality-assurance, and delivery scope around an agreed schema and review process. See the Annotation and Labelling Services page.
Can the collection cover a particular African market or setting?
Tell us the market, setting, objects, contributors, capture conditions, and delivery requirement. We assess the brief and confirm what is practical before making a commitment; this page does not claim universal market coverage.
How do image data and data-governance work relate?
Collection and governance decisions sit alongside the visual brief. We can make consent, provenance, buyer-diligence, data-flow, and safeguard questions visible in the project scope. See the AI Training Data Compliance page for the related operational support route.
Start with the visual task
Start an African image-data brief.
Send the outline you have. We will help turn the visual task, collection context, and quality requirements into the right next conversation.
A useful first email includes
- Intended model or product use and visual task
- Objects, scenes, documents, activities, or other categories to include
- Markets, settings, contributors, and capture conditions, if relevant
- Target volume, image format, metadata, labels, and quality requirements
- Rules for devices, angles, lighting, backgrounds, variation, or exclusions
- Rights, consent, governance, delivery timing, and procurement context, if known
If you do not have all of this yet, send what you have. We will help clarify the remaining questions on a call.
Prefer to write directly? dataservices@bsgdataworks.com
WhatsApp only: +234 806 790 4903