Portfolio

industries: healthcare · retail · automotive · logistics · media

// projects

The work, project by project

Some of this work was delivered directly by DeeLab; some reflects projects our annotators completed before joining our workforce. Each entry says which.

Across image, video, audio, text and structured data, the same requirements apply every time: clear guidelines, consistency, and review at every stage.

3d_lidar 3D city-scene segmentation

45,000 city-scene images segmented for an autonomous vehicle company: vehicles, pedestrians, buildings and road signs labelled to train perception models for safer autonomous navigation. Work by annotators now on our team, in structured, reviewed workflows.

volume: 45,000 images

audio Baby-sound labelling

100 hours of baby sounds (cries, snoring and other vocalisations) labelled for a retailer’s product development and sound-detection work. Delivered by DeeLab under strict guidelines, with multi-stage reviews across the five months.

100 hours · 5 months · 4 annotators

audio Medical audio annotation

50 hours of bowel-sound recordings annotated for a healthcare facility, supporting AI research in gastrointestinal diagnostics. Delivered by DeeLab, with clear labelling instructions for the sound patterns and structured internal reviews throughout.

50 hours · 2 months · 5 annotators

image Aerial roof annotation

1,600 aerial roof images annotated with polygons for an architectural firm’s structural analysis, held to geometric accuracy standards with regular review cycles. Work by annotators in our network.

volume: 1,600 images

image Consignment image labelling

Bounding boxes and text capture from consignment images for a shipping company, delivered on tight deadlines to feed automated consignment tracking and logistics workflows. Work by annotators in our network.

video In-store video tracking

Customer movement and shelf picks tracked for a retail client: bounding boxes kept consistent across video frames to support behavioural analysis and store-layout decisions. Work by annotators in our network, inside a structured review workflow.

duration: 6 months

text Article classification

Articles from multiple media sources classified against editorial guidelines to assess suitability for publication, high-volume work that had to hold both speed and accuracy. Work by annotators in our talent pool.

text Review sentiment classification

Customer reviews for a beauty products company classified as positive, neutral or negative against predefined criteria, with regular quality checks: input data for the client’s sentiment analysis. Work by annotators in our network.

data_entry Data entry & bookkeeping

Daily customer payment records for a grocery store: transactions entered, reconciled and matched against receipts to keep financial tracking and reporting accurate. Work by members of our current workforce.

/training

Training & certification

We also train client teams, using the same practical curriculum our own annotators certify on at DeeLab Academy.

01 Data annotation training A two-month programme for 20 people at a BPO company, run by 4 trainers: image and video labelling, semantic segmentation and 3D LiDAR, with quality control as the through-line, hands-on exercises and feedback sessions throughout.

02 Image labelling essentials A one-week course for 7 trainees, run by 4 trainers: bounding boxes, keypoints and polygon labelling techniques, with practical assignments that prepared participants to start on live annotation tasks.

// tell us about your data

Send us a sample. We’ll send back a scoped quote in 48 hours.

Request a quote or write to [email protected]