HACARUS Launches HACARUS Field Vision Tuner, a Retraining Cloud App for Accident Prevention AI

HACARUS Inc. (Headquarters: Nakagyo Ward, Kyoto City; CEO: Takashi Someta) will launch "HACARUS Field Vision Tuner," a cloud application that allows users to retrain the PPE detection AI for its accident prevention AI solution "HACARUS Field Vision" to match their on-site environments, starting July 1, 2026.

"HACARUS Field Vision Tuner" is a cloud application that automatically extracts and trains learning images from videos uploaded by users. Users can optimize the PPE detection model to their specific site environments without needing specialized AI knowledge.

The generated trained models can be downloaded and applied to the edge AI boxes within the "HACARUS Field Vision" system. This allows for updating AI models in response to environmental changes while maintaining a configuration that does not require an internet connection for on-site operations.

UI Screen (Tagging Function)

Background of Development

The accuracy of PPE detection in manufacturing environments is heavily influenced by site-specific conditions. Factors such as the shape, color, and wearing position of PPE, as well as camera angles, lighting conditions, and worker movements, vary from site to site. Consequently, a general-purpose AI model may not achieve sufficient detection accuracy.

Previously, adapting to PPE not covered by standard functions or adjusting detection accuracy due to on-site environmental factors required retraining by HACARUS. This resulted in a certain lead time and cost for model adjustments whenever new PPE was introduced or the work environment changed.

"HACARUS Field Vision Tuner" supports more rapid updates of AI models in response to environmental changes at each site by enabling users to perform the retraining process themselves on the cloud.

Key Features of "HACARUS Field Vision Tuner"

Retraining Possible with Just Video Upload and Tagging By uploading videos filmed on-site and tagging the target PPE and conditions, the retraining process can proceed. Even personnel without specialized knowledge in AI or image processing can create models tailored to their on-site environment.

Automated Creation of Training Images The AI automatically extracts individuals from videos and creates training images for each detection area, such as the head, eyes, body, and arms. This streamlines the time-consuming training data creation process for retraining, reducing the user's workload.

Trained Models Can Be Applied to Edge AI Boxes Generated trained models can be downloaded and applied to the edge AI boxes of "HACARUS Field Vision." On-site operations can continue without an internet connection, allowing for model updates while maintaining existing security designs.

Supports Data Sharing and Permission Management by Organization Data can be shared within an organization, supporting operations by multiple personnel and data utilization between on-site and management departments. It also supports permission management, facilitating organized AI model operations.

With "HACARUS Field Vision Tuner," user companies can perform the creation of training data and model updates required for retraining PPE detection AI in shorter periods. This makes it easier to flexibly respond to the introduction of new PPE, changes in work environments, and differences in camera installation conditions, enabling the operation of PPE detection models tailored to the conditions of each site. Furthermore, by allowing users to perform retraining tasks themselves, it leads to reduced lead times and operational costs associated with external requests. This enables continuous adjustment and improvement of PPE detection AI according to the on-site environment, rather than a one-time implementation.

Future Outlook

HACARUS plans to expand "HACARUS Field Vision" into an on-site support AI solution that goes beyond accident prevention. By leveraging the person detection and behavior recognition technologies cultivated through PPE detection, HACARUS aims to broaden its applications to safety uses such as intrusion detection and fall detection, as well as areas like work analysis and productivity improvement, by expanding the range of detectable objects. "HACARUS Field Vision Tuner" will continue to be enhanced as a platform for customers to optimize newly expanded detection functions according to their on-site environments.

About HACARUS Inc.

With the mission "Providing next-generation 'measurement' for those who build the future," HACARUS offers "AI solutions" that lead to fundamental problem-solving with limited data. Since its establishment in 2014, the company has contributed to problem-solving for numerous businesses, leveraging its experience and unique technological capabilities to assetize human knowledge and achieve operational efficiency and labor savings. HACARUS provides integrated support from data acquisition to integration with existing systems, working towards a future where humans and AI coexist.

Headquarters: Oike Bldg. 8F, Takamiya-cho 206, Nakagyo-ku, Kyoto City, Kyoto Prefecture

Established: January 2014

CEO: Takashi Someta

URL: https://hacarus.com/ja/

Media Inquiries: pr@hacarus.com (Public Relations)