COVID-19 'Untact' Era: Contactless Image Analysis Becomes Important... Deep Learning Vision Software

Across industrial circles, 'untact' is emerging as a hot topic. 'Untact' is a Korean government policy/campaign word (un + contact) created to promote minimizing contact. At the same time, the importance of deep learning image analysis techniques that enable 'untact' is also increasing. Based on trained deep learning algorithms, different situations can be analyzed at a distance from images. Companies in various fields are rapidly adapting to the 'untact' era by using deep learning vision software.

In the insurance industry, artificial intelligence has been used services such as a system that automatically estimates the repair cost of a car by reading the car type, damage, etc. from just a picture of the accident vehicle, or a system determining the insurance value.

AI startup Neurocle is showcasing its deep learning vision software Neuro-T & Neuro-R, which can be universally adapted to the individual needs of various industries, offering ways to apply deep learning vision technology in the age of 'untact'.

■ Distribution & logistics — image analysis for unattended payments and efficient distribution

Recently, the distribution industry has been using smart cameras to introduce unmanned stores. With deep learning vision software, this is not a problem even in local stores. By creating a deep learning model trained with product images and classifying them with an intuitive GUI, it is possible to identify and classify items that customers want to purchase.

In the logistics industry, there is a high demand for automating packaging inspections. Ensuring that the right number or type of product is fully included in the package is one of the essential parts of logistics, but this process is done manually by human workers.

However, if images of various products are trained through deep learning vision software, the trained deep learning model will recognize the product in different environments, so that inspections can be automated.

■ Security – contactless image analysis for automatically judging situations

With deep learning vision technology, some companies provide services that recognize visitors or hazards with a camera at an entrance gate or in danger zones and analyze them against the trained information. Due to the 'untact' campaign, the number of people who want to assess the situation in a contactless environment is increasing, so the use of deep learning vision software is likely to be more frequent.

For example, various risk factors such as people or animals may appear near the train tracks, which need to be identified. External factors such as the background changes depending on the time and weather had previously made it difficult for machines to make consistent judgments.

△Security. Changes in background and optical conditions can be automatically recognized for situational analysis.

"We will be leveraging deep learning image analytics for a wider variety of industries," said Neurocle CEO Hongsuk Lee, "Deep learning image analysis technology requires deep learning experts, but it is more efficient to utilize generalized software that is both customizable for different businesses and usable for non-experts rather than spending unnecessary resources on learning the technology in depth." In response to the 'untact' era, Lee plans to incorporate this technology into business in various fields such as healthcare, education, insurance—anywhere that needs image analysis.

Neurocle will be participating in the KCR 2020 Virtual Congress on September 17, showcasing its deep learning vision software (Neuro-T & Neuro-R) and meeting customers from a variety of fields that require image interpretation for medical research.