Intenseye
Connect to the factory's existing surveillance cameras, using computer vision to watch for major workplace-safety risks 24 hours a day
Intenseye is a workplace-safety-tech company headquartered in New York, doing the work of connecting factories' and warehouses' existing CCTV footage into computer-vision models to automatically detect the dangerous behaviors and situations that 'once seen, should be stopped' — no hard hat, human-vehicle path crossing, entering restricted zones, poor posture, and so on — then notifying EHS (environment, health, and safety) staff in real time.
Features and application scenarios
This kind of product's core proposition is actually a math problem: a factory may have two or three hundred cameras but only a few safety staff, and human eyes simply can't watch them all. The official site claims the platform processes over 22 billion frames daily, and what it does is automate the 'watching.' What's special is its 3D human-pose tracking — without employees wearing any sensors, purely by image analysis of the torso, neck, and limb positions to judge whether a lifting posture will cause musculoskeletal injury, a practical angle in ergonomics. Besides pure software, they also launched a Sentinel hardware line, including IP cameras for industrial environments, thermal cameras, and audio-alert devices, for sites where existing surveillance coverage is insufficient. The official cited Swire Coca-Cola case claims a 60% reduction in hazards and a 27% drop in lost workdays.
Suited to manufacturing, logistics warehousing, food and beverage, and other sites with lots of on-site workers, especially companies that already have a bunch of surveillance cameras but only use them for after-the-fact retrieval. Taiwan's traditional manufacturing is actually the ideal customer for this kind of tool — enough camera density, high workplace-safety pressure (OSHA inspections and ESG reports both need numbers), but generally tight on-site safety staff. Note the privacy and labor-relations communication: turning cameras from 'after-the-fact verification' into 'real-time monitoring of employee behavior' is an issue in Taiwan's labor environment that needs serious handling, not something that's fine once installed.
Main features
- Connect to existing CCTV, using computer vision to detect workplace-safety violations and dangerous situations in real time
- 3D human-pose tracking, assessing lifting posture and ergonomic risk without wearables
- SIF (serious injury and fatality)-prevention-oriented event classification and prioritization
- Sentinel hardware line: industrial IP cameras, thermal cameras, traffic monitoring, and audio-alert devices
- Real-time alerts pushed to on-site and EHS managers
- Compliance-audit reports, usable for occupational-safety management and ESG disclosure
Common uses
- Factory on-site PPE (hard hat, goggles, vest) compliance auditing
- Real-time warning of forklift-and-pedestrian path conflict
- Detection of unauthorized entry into restricted, high-risk areas
- Ergonomic-risk assessment of lifting posture and work improvement
- Occupational-safety compliance reports and ESG safety-performance data output
Key Features
- Connect to existing CCTV, using computer vision to detect workplace-safety violations and dangerous situations in real time
- 3D human-pose tracking, assessing lifting posture and ergonomic risk without wearables
- SIF (serious injury and fatality)-prevention-oriented event classification and prioritization
- Sentinel hardware line: industrial IP cameras, thermal cameras, traffic monitoring, and audio-alert devices
- Real-time alerts pushed to on-site and EHS managers
- Compliance-audit reports, usable for occupational-safety management and ESG disclosure
Pros
- Can reuse existing surveillance cameras, with a far lower adoption barrier than a full hardware swap
- The wearable-free pose tracking is very practical — requiring workers to wear sensors daily usually has a terrible compliance rate
- Has named large-customer cases with quantified numbers, not just a concept demo
- Provides both software and industrial-grade hardware, covering surveillance blind spots
Cons
- The privacy dispute of employee monitoring is real, and pre-adoption labor-relations communication can't be skipped
- Pricing not public; it's an enterprise-grade project purchase
- Detection accuracy highly depends on camera angle and on-site lighting, so upfront tuning requires investment
- If false positives aren't tuned well, on-site staff will quickly become numb to alerts
Use Cases
- Factory on-site PPE (hard hat, goggles, vest) compliance auditing
- Real-time warning of forklift-and-pedestrian path conflict
- Detection of unauthorized entry into restricted, high-risk areas
- Ergonomic-risk assessment of lifting posture and work improvement
- Occupational-safety compliance reports and ESG safety-performance data output
Editor's Note
Editor's note: Honestly, the potential demand for this kind of tool in Taiwan's manufacturing is severely underestimated. Most factories' surveillance cameras only get pulled up after an incident, and by then the person is already hurt. Turning the same cameras into real-time warning is logically a worthwhile business. But I must honestly point out one thing: the biggest resistance to landing this in Taiwan isn't the technology or the budget — it's the on-site trust. If you go live without clearly explaining 'this is to protect you, not to catch you,' someone will block the lens with a cardboard box in the first month.
FAQ
Do I need to replace the existing surveillance cameras?
Not necessarily. Intenseye features connecting to existing CCTV, one of its main selling points. But if the site has surveillance blind spots or image quality is too poor, they also have the Sentinel hardware line to supplement.
Won't this have employee-privacy issues?
Yes, and it must be addressed head-on. This kind of system essentially analyzes employee behavior, so before adopting in Taiwan, communicating with the union or labor-management meeting, clearly defining the data purpose and retention period, and considering de-identification is advised. Technically feasible doesn't mean there's no compliance or labor-relations risk.
How do I choose among it, Protex AI, and Voxel?
The three are very close in positioning, all connecting to existing cameras to do workplace-safety computer vision. The difference is in details: Protex AI emphasizes an edge-computing privacy architecture, Voxel features 48-hour fast deployment, and Intenseye puts more into ergonomic-pose analysis and its own hardware. Directly having all three do a POC with your own site's real scenarios to compare is advised.
Related AI Tools
MOVUS
Australia's AI machine health monitoring that goes beyond alerts to tell you what to fix first
Luxonis
OAK series edge AI depth cameras that run computer vision directly on the device
Athena Security
Proactive concealed weapon detection using computer vision and walkthrough sensors—alerting security before a weapon is even drawn.
SoundThinking
An Operational AI Platform Unifying Gunshot Detection, License Plate Recognition, and Weapon Screening for Public Safety
Mark43
Cloud-Native Public Safety Platform for Law Enforcement and Dispatch with Built-in AI Reporting and Shift Briefings
Waste Robotics
AI-powered waste sorting robots utilizing hyperspectral imaging to detect material differences invisible to the naked eye.