- MapFan DB
Annotation outsourcing service
"TASUKI Annotation"
SoftBank Corp.
TASUKI Annotation: An annotation service that uses technology to create high-quality results.
Could you tell us about the overview of the service?
"TASUKI Annotation" is a service that handles the collection and processing of annotation data (training data and instructional data) necessary for AI model development. Using proprietary technology born from SoftBank's AI model development experience, it solves the problems of previous annotation outsourcing services.
Types of annotation include:
Image processing includes: image classification (tagging attributes), object detection (finding objects within an image and assigning meaningful tags), and region detection (identifying regions within an image and tagging each region).
• Natural language processing: Classification of texts and documents, sentiment analysis, etc.
• Video: Frame-by-frame cropping (optional) and annotation.
• Audio: From collecting audio data to transcribing audio.
This service is in demand across various industries and fields, including healthcare, manufacturing, drones, retail, and autonomous driving.

In the rapidly growing AI development market, annotation (creating training data) is an essential and time-consuming process. Factors contributing to the time-consuming nature of annotation include:
1. Because it is difficult to define requirements in advance.
2. Because it requires manual labor.
3. Because high-quality data is required.
However, "TASUKI Annotation" can solve these problems with the following approach.
1. Our "Project Registration System" assists in registering the requirements needed when submitting a project request, significantly reducing the workload involved in project requests and improving the quality of requests.
2. By using AI-powered automatic annotation, we significantly reduce work time and costs, and deliver the final product.
3. In addition to AI-powered automatic annotation, professional annotators meticulously check the final output and make corrections until the very last page. This ensures that we produce the high-quality data our customers require.
"TASUKI Annotation" provides a smart and efficient annotation service.

Please tell me about the features that use road driving image data, and their advantages and effects.
For autonomous driving annotation, we developed a dataset for AI development by annotating the vast amount of road driving image data from GeoTechnologies, Inc. Inc.
GeoTechnologies, Inc.' road driving images allow us to identify various facilities such as traffic regulations, road signs, signs, traffic lights, utility poles, and guardrails. This has enabled us to immediately provide data that can be used for AI development in all industries, including autonomous driving and road inspection.
Our service plans include "Detection," which detects objects and equipment from road driving images, and "Segmentation," which extracts them individually.

Please tell me about your implementation track record.
We have a proven track record of implementation in various fields, including illustration data annotation for AI mobile game apps that automatically recognize children's drawings, image data annotation for AI development that detects dangerous driving of construction machinery, and image data annotation for services that replace skilled trades with AI.
Could you tell us about the circumstances and deciding factors that led you to adopt our map products?
The speed at which negotiations progressed from the start of discussions to the final agreement was the deciding factor.
Could you tell us about your future plans?
We will gradually expand our repertoire of data through collaborations with companies like GeoTechnologies, Inc., so that developers can quickly obtain the data they need for AI development whenever they need it.
If you are having trouble with annotation or data collection, please contact us using the information below.
Click here for the "TASUKI Annotation" service website.
For inquiries about "TASUKI Annotation" please click here.