Bounding boxes
Tight 2D boxes around every object of interest, with class labels and attributes such as occlusion or truncation.
Object detection · vehicles · retail shelves
Annotationbd is a team of 20+ annotators led by Md Mamun Mia. We draw the boxes, polygons, masks and keypoints your computer vision models learn from, for clients in the US and worldwide.
We do annotation and nothing else. You bring the images and the guidelines; we return clean, consistent labels ready for your training pipeline.
Tight 2D boxes around every object of interest, with class labels and attributes such as occlusion or truncation.
Object detection · vehicles · retail shelves
Vertex-by-vertex outlines for irregular shapes, where a box would include too much background.
Rooftops · crops · damaged parts
Pixel-level masks so every pixel belongs to a class, including instance masks when objects touch.
Road scenes · medical imagery · satellite
Landmarks and skeletons placed in a fixed order, with visibility flags for hidden points.
Pose estimation · faces · hands
Text regions boxed and transcribed exactly as written, line by line or word by word.
Receipts · forms · signs · ID layouts
Whole-image or per-object tags against your label list, including multi-label and attribute tagging.
Content moderation · product catalogs
Best when every image needs one tag or a few objects.
Best when object counts vary a lot from image to image.
Best for pixel masks, changing guidelines or a dedicated team.
Rates depend on the annotation type, image complexity and volume. Send a few sample images and your guidelines, and we'll reply with a fixed quote.
You share your classes, guidelines and a handful of sample images.
We label a small batch for free review so you can check our reading of the rules.
The team works through the dataset, with edge cases logged and sent to you.
A second annotator checks each batch before it leaves our hands.
Labels arrive in your format, with fixes made if anything misses the guideline.
Annotationbd is run by Md Mamun Mia, who has spent more than five years labelling data for international companies. He built the team from people he trained himself, so the same standards apply to every image.
We're based in Bangladesh and work with clients in the United States and elsewhere. We take one-off batches and long-term or full-time annotation contracts.
We focus on annotation only. We don't train models or build software, which keeps our attention on label quality.
Excellent work! Very fast turnaround, great attention to detail, and followed complex annotation instructions perfectly. Communication was clear and professional throughout. Would definitely work with him again.
Excellent service from start to finish. He was professional, attentive to every detail, and went above and beyond to deliver an outstanding final result. Highly recommend!
Exactly what i asked for
5/5 Professional, detail-oriented, and easy to work with. Delivered quality work on time. Highly recommended.
Bounding box, polygon, semantic and instance segmentation, keypoint, OCR and text, and image classification annotation. We focus on annotation only and do not train models or build software.
Email a few sample images and your label list to annotationbd@gmail.com, or order through our Fiverr gig or Upwork profile. We start with a small pilot batch so you can check our work before the full dataset.
It depends on the annotation type, image complexity and volume. We price per image, per object or per hour. Send a few sample images and your guidelines and we will reply with a quote.
COCO JSON, YOLO TXT, Pascal VOC XML, CSV or your own custom schema.
More than 20 trained annotators, led by Md Mamun Mia, who has more than five years of experience labelling data for international companies.
Yes. We take one-off batches as well as long-term and full-time annotation contracts.
We are based in Bangladesh and work with clients in the United States and around the world.
Email us a few sample images and your label list. We'll send back a quote and can start with a pilot batch.