Vesta Inc.

Predictive Modeling and Analytics

Our Mission

with Momentum

Our mission is to prove that the impact of all climatic and weather-related natural disasters can be predicted.
We predict the occurrence of climatic and weather-related natural disasters and the estimation of loss caused by them. Once the predictive analytics become advanced with developments in research, evolution of technology, and the internet of things, these events will eventually become predictable. We shall continue to release products that will build the foundation for achieving this at the earliest. The name Vesta is derived from the terms "prove" and “predictable.”

Our Business

service icon

In-house R&D (B to C)

We create in-house apps on weather and disasters for our end users. We pursue innovative functions, usability, and the timely release of new products. The main processes we use are hypothesis testing and interviews.

service icon

Contract-based R&D (B to B)

We deliver models and applications to our clients after conducting thorough academic research on weather and disasters. We have rich experience in developing products for the Amazon Web Services cloud. We also provide extensive knowledge of coding and implementation of applications.

Vesta News

Update cmap.dev, the real-time loss prediction website, with Aioi Nissay Dowa Insurance Co., Ltd. The new function is to display the flood risk level in 1-km mesh in real time.  see Newsletter

- Takeshi Okazaki, Vesta Inc. 26 August 2020

Update cmap.dev, the real-time loss prediction website, with Aioi Nissay Dowa Insurance Co., Ltd. The new function is to estimate the loss before a typhoon landfall.  see Newsletter

- Takeshi Okazaki, Vesta Inc. 5 June 2020

Uploaded the photos of the field survey of Typhoon Hagibis (No.19) in 2019 to Blog.  see Blog

- Takeshi Okazaki, Vesta Inc. 14 October 2019

Uploaded the photos of the field survey of Typhoon Faxai (No.15) in 2019 to Blog.  see Blog

- Takeshi Okazaki, Vesta Inc. 15 September 2019

Blog

Publications

Proposal of a method to estimate insurance claims caused by heavy rain,
The 33rd Annual Conference of the Japanese Society for Artificial Intelligence, 2019

In order to facilitate flood and landslide damage assessment, we propose an evaluation method that can predict the total number of damaged houses with high accuracy immediately after a disaster. To improve accuracy, we narrowed the exposure down by using low-lying and landslide high-risk areas and apply the precipitation of 15-year return period as a threshold to the original precipitation to take account of disaster preparedness by each local government. When applied to two heavy rain events occurred in 2017 and 2018 and validated the method, we confirmed that it can be well reproduced in any prefecture.

see publication
Real-time control of unmanned aerial vehicle by object detection for loss assessment,
The 32nd Annual Conference of the Japanese Society for Artificial Intelligence, 2018

During a loss assessment after a disaster, a drone is useful to take the photos of the building roof. However, there is a possibility of touching with obstacles and persons when using drone. Therefore, in this study, we developed the system that can control drone in real-time by combining the object detection called YOLOv2 with the drone control program. By outdoor test flights, we confirmed that the system can recognize the person's position and control the drone’s movement according to the position.

see publication
Identifying building characteristics by deep learning and a method of assessing typhoon losses,
The 31st Annual Conference of the Japanese Society for Artificial Intelligence, 2017

In order to facilitate typhoon damage assessment, we propose an evaluation method that can predict the total number of damaged houses with high accuracy immediately after a disaster. The exposure was developed in building level and the buildings characteristics were taken into consideration such as construction year, roof shape, and roof area. Since the roof related information was not collected in any industry, analyzed by deep learning applied to aerial imageries. When applied to three typhoons in 2015 and validated the method, we confirmed that it can be well reproduced in any prefecture.

see publication
Application of the Typhoon Model for the Non-life Insurance Industry,
Japan Association for Wind Engineering (JAWE), Volume 41 (2016) Issue 2, pp. 152-160

The loss caused by typhoons and floods are generally borne by non-life insurance companies instead of individuals. These companies retain a large quantum of accumulated risk. Current efforts by the non-life insurance industry form an important aspect of reflecting on the history of catastrophe modeling and wind engineering in Japan. The industry uses a typhoon simulation model based on the Monte Carlo simulation for risk management. This paper presents its historical background, provides an overview of the typhoon model, and posits future challenges.

see publication
Analyzing All Roofs Throughout Japan to Estimate Loss of Buildings,
Risktaisaku.com Journal

A new website called cmap.dev assesses in real time the number of damaged buildings for each municipality at the time of a disaster. We interviewed Dr. Takeshi Okazaki regarding his views about this first-ever attempt of its kind.

see publication

Questions and Answers

Who are Vesta’s team members?

The members of our team are highly qualified professionals in weather, disasters, and disaster prevention and are very experienced academic researchers as well. We also collaborate with highly skilled app development engineers and designers.

Where is Vesta headquartered?

Our office is located in Plug and Play (PnP) in Shibuya, Tokyo, Japan. PnP is a co-working space.

Do you offer contract-based R&D?

Yes, we do. We take a wide range of approaches on contract-based development, including extensive knowledge, and we propose products that exert a positive impact on society. Most of our clients are non-life insurance companies, but orders from other industries are welcome. Please contact us at support.tm@vestapd.com

What are Vesta’s unique characteristics?

Many companies can develop apps and have the expertise to analyze the weather and predict disasters. However, only a few companies have the skill and ability to develop new helpful products using updated technology to predict, calculate and mitigate losses. We differentiate ourselves from other companies by offering untapped services.