
Interview Cooperation: Business Design Group Headquarters Business Development Department Mr. Keita Kawakami
SCSK Corporation Overview
・Established: October 25, 1969
・Capital: 21,285 million yen
・Number of Employees: 15,328 (consolidated), 8,470 (individual) (as of March 31, 2023)
Business Overview: SCSK Corporation offers a full lineup of all IT services necessary for business, including consulting, system development, verification services, IT infrastructure construction, IT management, IT hardware and software sales, and BPO (Business Process Outsourcing).
Services Utilized
- Annotation Services
Challenges Before Implementation
HS: Please tell us about the project and outsourcing details you requested from our company.
Mr. Kawakami: This is a project to demonstrate anomaly detection using AI, aimed at improving the efficiency of maintenance operations that are currently managed manually in a certain industry.
The contents of the commission are as follows:
・Annotation (Bounding Box / Semantic Segmentation)
・Pickup (Selecting images that contain the target object)
・Generation of difficult-to-collect abnormal images (Generating synthetic images using image generation AI, image editing software, etc.)
HS: If you have any concerns or challenges while working on the project, please let us know.
Mr. Kawakami: Yes, there were several issues. First, last year, half of the annotation work was done in-house, and the other half was outsourced to external companies.
Increase in Time Spent on Annotation
The first issue is the labor shortage. Annotation work requires a significant amount of effort and time, making it difficult for our team to handle the workload.
Improving AI accuracy by enhancing the quality of annotations
The second point is that the previous system did not have annotation experts, making it difficult to establish and refine annotation standards that would yield consistent results regardless of who performed the task. This led to variability in the quality of annotation deliverables. The low quality of annotations was one of the reasons why AI accuracy did not improve.
Reasons for Choosing Human Science
HS: Could you tell us what prompted you to request HS?
Mr. Kawakami: Yes. In order to address the challenges we have faced so far, we began considering outsourcing to annotation companies with expertise. However, at that time, the companies performing annotation work were not registered as our outsourcing partners.
Therefore, we started by investigating companies that conduct annotation. Initially, we picked about 60 companies and requested materials. At this stage, we narrowed it down to about 40 companies based on factors such as management status, financial aspects, credibility, industry share, future potential, and risks, and then established 20 evaluation criteria to assign points, ultimately narrowing it down to 10 final candidates.
After that, we directly contacted these selected companies, conducted detailed hearings and annotation trials, and ultimately narrowed it down to three companies, from which we chose your company.
HS: Please tell us the main reason for choosing our company.
Mr. Kawakami: One of the most important aspects was to find a company that could build a long-term business relationship. Not only for this project, but also because there is a possibility of outsourcing annotation work from various departments at SCSK in the future, it was necessary to select a company that has the capability to handle multiple projects and large-scale projects.
Implementation Effects and Evaluation of Human Science Co., Ltd.
HS: Were there any effects obtained after the implementation?
Mr. Kawakami: This reflects the reasons for requesting your company and our expectations of you, and it has been effective in the following ways.
<1. Focusing on AI Development Tasks>
When we were doing annotation work in-house, it took a lot of time. It was difficult to establish annotation standards, leading to multiple revisions and unnecessary work. By outsourcing the annotation tasks, we were able to create an environment where we could focus on AI development. This allowed us to concentrate the necessary resources on development, improving development efficiency. Being freed from annotation work enabled us to prioritize development more effectively.
<2. Higher quality annotation deliverables obtained than in-house production>
Regarding the quality of annotations, we expected high-quality deliverables that were difficult to achieve in-house. By consolidating the Q&A and feedback between your company and ours during the work based on the prepared standards document, the criteria became clearer. As a result, the quality of the annotations improved compared to in-house production, significantly contributing to the accuracy of the AI.
<3. Achieving Shortened Development Periods Through Division of Labor>
Since the number of engineers we can secure in-house is limited, it was a challenging task to conduct annotation work and AI development simultaneously without sufficient time for development. However, by outsourcing the annotation work to your company and dividing the labor, we were able to focus on the AI model development and accuracy improvement, successfully completing AI-related projects in a short period of time.
<4. Acquisition of Knowledge Regarding Annotation Work>
We gained a lot from your company regarding the sharing of annotation knowledge. You taught us specific methods for annotation work, perspectives to consider while performing the work, and tools used, through our repeated meetings. In particular, this time, you provided demonstrations and lectures on specific segmentation methods and concrete ways to improve work efficiency, which contributed to enhancing our knowledge. We appreciate your company for these aspects as well.
<5.Cost Reduction>
Finally, we were able to gain benefits in terms of cost, contributing to the project's profitability. Overall, we evaluate that outsourcing the annotation work has had a positive impact in many aspects.
HS: Please share your experience with our company and your evaluation from the time of implementation to the present.
Mr. Kawakami: To be honest, I have hardly any issues with your company and am currently only satisfied. I have a very good impression. Regarding annotations, I would like to continue requesting your services as much as possible, as long as the budget allows.
Future Prospects
HS: Please let us know if you have any images or prospects for utilizing our services in the future.
Mr. Kawakami: I feel that with the increasing and improving AI technologies in society, our company will also need to utilize AI technologies more in the future. In that regard, I would like to maintain a long-term relationship with your company.
Currently, each department is developing AI, but we are not aware that we can outsource annotation work to your company, and there is a possibility that we are doing it in-house. Therefore, I believe we should actively promote and communicate the many benefits of using your annotation services for the entire SCSK.
HS: We are pleased that you chose our company to achieve results more efficiently. We value building long-term partnerships with our clients and walking alongside them. We are proud that your vendor selection aligns with your values. If there is anything we can assist you with in the future, please feel free to reach out. We look forward to working with you.