Collecting Photos of Bald Men for a Medical Client

We found a fast and efficient solution for gathering 200 sets of photos of bald men from different angles. Thanks to a detailed guideline, we annotated the images according to the Norwood scale with the highest accuracy.

Industry:
Medicine
Data:
200 annotated sets
Timeline:
1.5 months
Industry:
Medicine
Data:
200 annotated sets
Timeline:
1.5 months

The Task

Imagine this: A medical company conducting advanced research on baldness approaches us with an unusual request.

They needed a dataset of bald men. Each set had to consist of five photos of the same person taken from different angles: front view, profile (both sides), top view, and back view.

But it wasn’t just about collecting photos—it was about creating a database where each image was “labeled” with medical precision. The reference point was the Norwood scale, a classification system for the stages of baldness. There was no room for error.

The Solution

  • 01

    Preparation Stage:

    • Studying the Norwood Scale: We dived deep into the topic, researching everything available about the scale.
    • Guideline Creation: With the help of medical experts, we developed an extremely detailed guide for our annotators, describing each stage of baldness down to the millimeter, complete with illustrations and diagrams.
  • 02

    Annotator Training and Preparation:

    • Every annotator had to pass an entry test on the Norwood scale (a 50-example test) before starting work.
    • We conducted hands-on training on annotating photos with different degrees of baldness, analyzing ambiguous cases, and discussing evaluation criteria.
    • After training, we calibrated the annotators’ work to ensure uniformity and precision.
    • Annotators were also trained in using spreadsheets, data processing platforms, and annotation tools.
    • To support the team, we set up a helpdesk for real-time expert assistance.
  • 03

    Data Collection:

    • A major crowd-platform served as the primary channel for collecting photos.
    • We created a clear and engaging task for participants, complete with detailed instructions and examples. To encourage high-quality submissions, we offered competitive compensation.
    • Our team closely monitored the process to promptly identify and address potential issues.
  • 04

    Data Validation:

    • We implemented a multi-layered verification system. First, an automated check ensured that all required photos were present and met technical specifications.
    • Then, experts manually annotated the images using the Norwood scale, cross-referencing them with our guidelines.
    • To guarantee accuracy, we conducted a cross-validation, where one set of experts reviewed the work of others.

The Result

  • Not only did we meet the deadline, but we also achieved near-perfect annotation accuracy. Thanks to thorough annotator training and a multi-step validation process, annotation errors were kept to a minimum—less than 1%.

  • The client approved all the data and was highly satisfied with the results.

Our Cases

Case studies highlight how our services have enhanced AI model training and improved business outcomes across various industries See more
  • Data Collection for Anti-Spoofing Tasks

    Within a month, more than 10% of the entire database with over 2,000 photographs was collected

  • Data Collection for Facial and Speech Recognition

    Through data collection, the client improved their biometric system for facial and voice recognition by 21%.

  • Content Moderation on the Video

    Content moderation on the video hosting platform enabled a 99% reduction in the influx of prohibited content

Image for form
logo
Andrey,
Head of Sales

Ready to work with us?

    This website uses cookies to enhance your experience, analyze traffic, and deliver personalized content and ads. By clicking "Accept", you consent to the use of cookies, as described in our Cookie Policy. Please choose your cookie preference.