Visir & AJ Hackett Bungy New Zealand
Vision AI Proof of Concept
Enhancing safety and operational efficiency using computer vision and machine learning in high-adrenaline environments.
Executive summary
"Safety is one of AJ Hackett Bungy New Zealand’s core values and is deeply embedded in daily operations. Their crew operates in controlled high‑risk environments where precise processes, double‑checks, and behavioural consistency are critical. Proactive Monitoring—monthly structured review of crew actions via CCTV—helps maintain high standards but is time‑intensive for senior leaders."
AJ Hackett Bungy New Zealand partnered with Visir to explore how computer vision and machine learning could enhance proactive safety monitoring across their high-adrenaline operational environments. The proof of concept (POC) demonstrated that modern vision AI could reliably detect crew, customer movements, and key safety behaviours, providing a foundation for future real-time monitoring solutions.
The Challenge
Defining the path to automated safety surveillance.
Reduce Review Time
Manually reviewing hours of CCTV footage for safety compliance is labor-intensive and prone to human oversight.
Strengthen Standards
Implement a fail-safe digital 'second set of eyes' to ensure rigorous safety protocols are followed during every jump.
AI Maturity Test
Evaluate if current Vision AI technology is mature enough to handle the complex, fast-moving environment of bungy jumping.
A core question emerged:
Can machine learning detect key crew safety behaviours with accuracy and consistency?
POC Objectives
The Proof of Concept focused on five critical detection capabilities to validate the system's effectiveness in real-world conditions. The goal was to determine feasibility, accuracy, and potential value before investing in a production‑scale system.
- 01Jump Cycle Identification
- 02Deck Entry Detection
- 04Crew Member Face Detection
- 03Tracking Virtual Line Crossings
- 05Carabiner Identification & Placement
Visir's Iterative Approach
Building high-stakes AI requires more than just technical skill; it requires a deep understanding of the operational reality. Our approach centered on collaborative design sessions with the Bungy NZ safety team.
Visir worked iteratively with AJ Hackett Bungy New Zealand through design sessions, workshops, prototype walkthroughs, and frequent communication. Requirements evolved as both teams learned more about operational realities—especially the complexity of systemising processes that humans typically navigate intuitively.
We utilized an agile, iterative workflow to manage scope effectively, moving from raw video data to refined neural networks. This ensured that every “detection” was tuned to the specific environmental nuances of the site while managing scope, budget, and proof‑level constraints.
The Solution: Intelligent Web Application
A central hub for safety intelligence and vision analysis.
Visir delivered a functional web application capable of ingesting and analysing site footage, detecting jump cycles, recognising crew, and identifying carabiners on deck.
Footage Ingestion
Secure analysis of existing CCTV streams.
Cycle Detection
Detection of jump cycles.
Perimeter Tracking
Instant alerts for virtual line crossing violations.
Face Recognition
Identifying Crew for access logs.
Carabiner ID
Fine-grained detection of equipment integrity.
The POC confirmed that Vision AI could detect carabiners, crew, and movements with strong accuracy. While operational complexity required deeper process thinking, the concept was proven sufficiently to justify further investment. Stakeholders across management—operations, H&S, engineering, and executive leadership—were aligned and supportive.
“We engaged Visir to work with us on a proof of concept to test vision and machine learning for detecting operational processes, with the objective of saving time and ensuring our people and customers always operate safely. We were able to prove the concept and were happy with the engagement, delivery, and professionalism throughout the project.”
Lessons Learnt
Operational Variability
Weather and lighting at the bridge site significantly impact vision accuracy, necessitating robust data augmentation. There is always more operational variability than expected when systemising processes.
Depth vs Cost
Proof‑of‑concept work requires balancing depth, cost, time, and clarity of outcomes. Early and continuous stakeholder involvement helps align expectations.
Proof‑of‑concept work requires balancing depth, cost, time, and clarity of outcomes. The UI was necessary to demonstrate success, even though modelling was the primary goal.
Next Steps
Ready to transform your business with intelligence?
Join industry leaders in implementing world-class vision AI solutions.
Get Started Now-p-500.png)
