Deep Learning Image Recognition Platform "Cloud" Is Now Available
- The new cloud service to validate the effectiveness of AI image recognition technologies for completion inspections of newly built condominiums in cooperation with Frontier Construction & Partners Co., Ltd. -
NTT Comware Corporation (Head office: Minato-ku, Tokyo, President: Satoshi Kurishima, hereinafter, NTT Comware) launches the sale of the "cloud version" of Deep Learning Image Recognition Platformv*1 on June 30, 2017. The cloud platform engages new add-on functions. We will further promote AI technologies to use in the various fields of "monitoring/censorship", "maintenance/inspection" and "product inspection" along with the PC version, which is already on sale.
Furthermore, NTT Comware starts a collaborative project with Frontier Construction & Partners Co., Ltd. (Head office: Minato-ku, Tokyo, President: Tsutomu Matsumura, hereinafter, FC&P*2) to conduct demonstration tests of AI technologies for completion inspections of their newly built condominiums by using the Deep Learning Image Recognition Platform "Cloud" from July 2017.
* Deep Learning Image Recognition Platform is a platform using AI technologies commonly used for various NTT Comware technologies*3 for NTT Group's AI "corevo®" *4.
In recent years, the importance of monitoring or "eyes of experts" is increasing in various industries and areas. The purposes include: to eliminate inappropriate content on the Internet, to detect any abnormalities on images from detection cameras, to diagnose deterioration of infrastructure such as buildings, tunnels and roads, and to find defective products in production lines that could rarely occur. Amongst, they have common issues in these situations; "lack of manpower/cost increase due to the need for quality information or its volume for each task", or "lack of skilled human resources to inherit expertise in each field due to aging of experts".
Under these circumstances, AI is expected to be a solution to these problems. NTT Comware has been promoting AI technologies to use in business, with the Deep Learning Image Recognition Platform "PC Version", which was released on March 1, 2017. By introducing the "cloud version", it is expected to make AI easier for customers to use in the fields of "monitoring/censorship", "maintenance/inspection" and "product inspection".
The following are the features of the services of "Deep Learning Image Recognition Platform" Cloud.
(1) Quick start to use AI technology
No software installation needed. It is easy to create AI by having it learn images.
(2) Less initial cost for AI
All you need is a license fee to use AI. No extra fee is needed for hardware/software purchase costs, labor costs for operation, office rent or electricity.
(3) Flexible to change the number of usages
Capacities of AI can be changed as required, to meet whatever customer needs, such as the following cases: expansion of business or change of business domains. The usage pattern can be changed easily for some types of businesses that the volume may be changed depending on busy/off-seasons: more licenses for the peak season or suspension of the service for off-seasons.
(4) Onsite AI package is available for engineers or staff members
For "maintenance/inspection", experts may often need to visit the site to check the target structures. They even need to shoot some photos and check the results of inspections onsite. If they have pre-equipped cameras in devices that they use daily, such as smartphones or tablets, they can evaluate the status on the spot. This is a revolution in routine tasks.
Cloud version: June 30, 2017 (Fri)
(The PC version has been available since March 1, 2017)
FC&P receives a number of requests for their quality inspection services*5, from developers and construction companies in new condominium construction. The FC&P engineers with high expertise and skills and their experienced staff members conduct inspections. In this demonstration test, NTT Comware's Deep Learning Image Recognition Platform "Cloud" is applied to the finish inspections (to check items including scratches, dirt, or any other defects on the finished surface), which is one of the inspections to conduct at the end of construction of the newly built condominiums of FC&P. The two companies conduct these tests in cooperation to verify its effectiveness. First, FC&P provides images of scratches, dirt, and other defects of the finished surface. NTT Comware then creates an image recognition engine by having it learn the image group. Based on the images or information, AI combines the pre-learnt images and the new images of the finished surface provided, so that the test is to see how precise AI can do the test, compared to the inspections by the engineers.
NTT Comware and FC&P will use the data from the test and study the applicability of the service of Deep Learning Image Recognition Platform "Cloud" to the finish inspections of real estate properties.