Page 107 - AI for Good-Innovate for Impact Final Report 2024

AI for Good-Innovate for Impact

Use Case – 23: Generative AI based Video Analytics for Improving

Worker Safety 23-Securade

Country: Singapore

Organization: Okyasoft Pte Ltd

Contact person: Asankhaya Sharma, asankhaya@ securade .ai

23�1� Use case summary table

Domain Safety monitoring
Problem to be Lead time and effort needed to create models for safety monitoring
addressed for enterprises (logistics and manufacture)

Fine tuned models for Video analytics at the edge with real time
Key aspects of the monitoring (learnt from visual prompting and images from existing
solution
cctv cam)

Zero shot object detection
Technology keywords
Visual prompting
Data availability private data

Metadata (type of Video from existing cctc cam
data)

Model Training and 0 shot, Conventional model is finetuned with data generated from
fine tuning gen AI model (like distillation)

Feature in the WSHAsia Magazine: link
Testbeds or pilot Vestar Iron Case Study: link
deployments
SDTA-Lenovo-Securade.ai partnership: link

23�2� Use case description

23�2�1 Description

Securade.ai's innovative Generative AI-based Video Analytics platform is revolutionizing the
field of worker safety in high-risk industries such as construction, manufacturing, and logistics.
Traditional safety monitoring systems fall short due to static rules, delayed real-time alerts, and
a lack of customization, often leading to inadequate workplace safety. The advent of generative
AI has introduced a paradigm shift, enabling the transformation of conventional CCTV cameras
into intelligent safety monitors. These AI-enhanced systems can provide real-time alerts for
safety violations, ensuring a proactive approach to accident prevention and enhancing the
overall safety culture within workplaces.

91