By clicking “Accept”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.
May 16, 2024

Use case: look back at our partnership with IBM

May 16, 2024

Use case: look back at our partnership with IBM

For 110 years, IBM has been the partner of choice for businesses, offering the most comprehensive range of IT services and financing to help them differentiate themselves in the marketplace. 

Also a leader in the field of AI & IoT thanks to their massive investments in research and quantum computing, IBM is now developing solutions for predictive maintenance that integrate computer vision models for automated visual inspection.

For the past year, isahit has been working with IBM on training these algorithms, helping them obtain optimal quality annotations while respecting urgent deadlines so that the algorithms are as accurate as possible.

IBM in a nutshell

  • Created in 1911 
  • A presence in 170 countries 
  • 350k employees worldwide 
  • World's leading company in terms of monthly patents filed since 1944

Isahit in a nutshell: 

  • First ethical data labeling platform for AI and data processing in Europe
  • 2000 projects with over 1700 HITers (our community of women working around the world)
  • HITers in over 37 countries
  • 250+ satisfied customers


Maximo Visual Inspection, a predictive maintenance tool from IBM

Maximo is a tool that uses computer vision technology to automate the visual inspection of buildings, infrastructures or production tools. Depending on the needs of their customers, IBM has to refine the training of its models so that they perfectly fit the use case.


The challenges faced by IBM

Learning a visual recognition model is a very long process: 

  • a large volume of annotations is required for a model to recognise an item with high accuracy.
  • there are a multitude of items to recognise 
  • there are also many iteration phases before full-scale labelling


Learning the Maximo Visual Inspection model requires significant human and technological resources to carry out this work. 


Isahit added value on the Maximo Visual Inspection Tool

Isahit appeared to be the most suitable solution as it offers :

  • A 100% customisable data annotation platform, 
  • A diverse workforce: Isahit has a presence in over 37 countries covering a wide linguistic, cultural and technical range. 
  • A qualified workforce, trained by Isahit's teams and selected according to the needs of the project.
  • Enhanced quality control processes: pre-project testing & review of annotations throughout the project.


The processes put in place for the successful completion of the project

In order to ensure that the project is carried out correctly, the IBM and isahit teams have set up specific and precise processes. 

IBM ensured that the following elements were prepared in advance of the project/launch: 

  1. Selection of images to be annotated
  2. Data sharing
  3. Setting targets: qualitative and quantitative

The isahit team assigned the project to a dedicated manager and implemented the following processes: 

  1. Selection of HITers according to their skills and the estimated volume
  2. Training of female contributors and management of skill development
  3. Test phase for validation of quality objectives
  4. Production phase (mass annotation of data)
  5. Deadline management and SLAs


IBM feedbacks: why choose isahit ?

  • Flexibility: adapting resources to real time needs
  • Reactivity and workforce to meet urgent deadlines
  • Qualified HITers, selected on the basis of their skills
  • Extremely demanding quality control
  • A business solution combined with a social approach to generate social impact
  • IBM & ISAHIT share the same values: to encourage social reintegration through digital inclusion and to support disadvantaged communities


A double victory: tech & social

One year after the beginning of our collaboration, we are mutually very happy to be able to make the following assessment

  • A project implemented in 24 hours
  • A re-trained model in 48 hours
  • 100% of annotations validated 

You might also like
this new related posts

Want to scale up your data labeling projects
and do it ethically? 

We have a wide range of solutions and tools that will help you train your algorithms. Click below to learn more!