Image Segmentation Annotation
for Pixel-Level Precision.
Semantic and instance segmentation for computer vision training datasets. Polygon annotation, per-class quality checks, and IAA measurement on every batch.
What Is Image Segmentation?
Image segmentation is the task of partitioning an image into meaningful regions by assigning a label to every pixel. Unlike bounding box annotation, which draws a rectangle around an object, segmentation follows the exact boundary of the object — making it significantly more precise and significantly more demanding to annotate correctly.
Segmentation datasets are used to train models that need pixel-level understanding of a scene — medical imaging systems that must identify exact tissue boundaries, autonomous vehicle systems that must distinguish road from pavement from sidewalk at the pixel level, or robotic systems that must calculate the exact shape of an object to grasp it correctly.
The quality challenge in segmentation annotation is boundary accuracy and consistency. Two annotators drawing the boundary of the same object will produce slightly different polygons. Aeterna Data measures this variation using inter-annotator agreement and enforces per-class boundary rules to keep the variation within acceptable limits.
Semantic vs Instance Segmentation
Semantic Segmentation
Every pixel in the image is assigned a class label. All pixels belonging to the same class — for example, all road pixels, all sky pixels, all building pixels — receive the same label. Individual instances of the same class are not distinguished.
Use when: Your model needs to understand what type of region each pixel belongs to — driving scenes, satellite imagery, medical tissue classification.
Instance Segmentation
Each individual object instance receives its own unique mask, even when multiple instances of the same class appear in the image. Three cars in an image produce three separate masks, each labeled as 'car' but individually distinguished.
Use when: Your model needs to count, track, or individually interact with objects — robotics, counting systems, multi-object tracking.
Use Cases
Autonomous Driving
Road, lane, vehicle, pedestrian, and sign segmentation for scene understanding in self-driving systems.
Medical Imaging
Organ, lesion, and tissue boundary delineation in radiology, pathology, and surgical planning datasets.
Satellite & Aerial
Land use, building footprint, and vegetation segmentation in aerial and satellite imagery.
Robotics
Object shape and boundary extraction for robotic grasping, manipulation, and scene understanding.
Agriculture
Crop, weed, and disease region identification in drone imagery for precision agriculture systems.
Industrial Inspection
Defect boundary delineation in manufacturing quality control and surface inspection datasets.
Annotation Method
Polygon Annotation
Annotators draw precise polygon boundaries around each object by placing vertices along the object edge. The polygon is then filled to produce the pixel mask. Polygon annotation gives the highest boundary accuracy and is the standard method for training-grade segmentation datasets.
Per-Class Boundary Rules
Each object class in the dataset has specific boundary rules defined in the annotation guidelines — for example, whether to include the shadow of an object, whether to annotate partially occluded objects, and how to handle object boundaries that overlap. These rules are validated during the pilot phase before production begins.
Occlusion Handling
Objects that are partially hidden behind other objects require explicit handling rules. Aeterna Data documents the occlusion policy for each class during guideline development — whether to annotate only the visible portion, or to infer and complete the full object boundary.
Quality Standard
Every segmentation annotation batch delivered by Aeterna Data is measured for inter-annotator agreement before delivery. IAA is not a target — it is a threshold. Batches that do not meet the threshold are reworked before the client receives them.
Cohen's Kappa
Pairwise IAA between two annotators on the same image set. Applied on every batch.
Fleiss' Kappa
Multi-annotator IAA across three or more annotators. Applied on complex multi-class tasks.
Every batch delivery includes a quality report: IAA scores by class, box count, annotator distribution, and any deviation notes. No dataset delivered without the quality documentation.
Our Workflow
Project Brief & Scoping
You share your dataset sample, object classes, and output format requirements. We scope the pilot — sample size, timeline, and deliverables — and confirm before work begins.
Pilot Phase
We annotate a representative sample covering the key object classes and scene types in your dataset. Boundary accuracy and IAA are measured. Guideline gaps are identified and resolved before production annotation begins.
Production Annotation
Full dataset annotation following the validated workflow. IAA measured on every batch. Rework at no cost for any batch below threshold.
Delivery & Documentation
Annotated dataset delivered in your specified format with a complete quality report — IAA scores, class distribution, and annotator notes.
Annotation Tools
Aeterna Data works with the annotation platform your team already uses. We do not require you to adopt a specific tool. If you do not have a platform, we can advise on setup based on your task type and scale.
How to Start
Send a Project Brief
Share your dataset sample, object classes requiring segmentation, semantic or instance type, and approximate image count. Use the contact form or email [email protected].
Send BriefReceive a Pilot Proposal
We send a scoped pilot proposal within 48 hours — sample size, timeline, flat-rate pilot fee, and deliverables.
NDA and DPA Signed
Before any data is shared, NDA and Data Processing Agreement are signed. Your dataset stays confidential.
Pilot Begins
Annotation starts inside your platform. IAA report delivered with the pilot dataset. Production follows on your confirmation.
Ready to Build Your Segmentation Dataset?
Start with a flat-rate pilot. Boundary rules validated. IAA measured. Quality report on delivery.