Paphavee

UX projects

→ Avigilon
→ Calipsa
→ CoinMarketCap
→ Digital Lending Chatbot
→ Mobile Wallet
→ In-store Companion
→ Motion design
→ Socialgiver
→ Ticketmelon

Fun projects

→ The Purist Artisan
→ Happy New Year!
→ These are Pieces of Life
→ Double Trouble
→ Did you know 👀 ?

CV 📄
Mail 📫

Product Design

Advanced rules

Jan 2021 - Present

Team: 1 Product Manager / 1 full-stack engineer / 1 designer
Status: Live in 2021, still constantly iterating.

Beyond the Basics:
Addressing Complexities in CCTV Detection 💡

Calipsa's foundation product has revolutionised CCTV analysis with its impressive accuracy in object detection. However, the security landscape is ever-evolving, and simple human and vehicle detection may not suffice in all situations. Here's where the challenge lies:

Complex Scenarios: Real-world environments present complex situations with multiple objects, obscured views, and unusual activity patterns. Calipsa needs to adapt to these complexities to maintain its effectiveness.
Specificity Demands: Investigations often require pinpointing specific actions or interactions beyond basic object presence. The current system might miss crucial details like suspicious behavior or object transfer.
False Positives: While the current accuracy rate is high, even a small percentage of false positives can waste valuable investigative time on irrelevant footage.

Solution:
Unlocking Advanced Detection with Customisable Rules 🔧

To address these challenges and expand Calipsa's capabilities, we propose a solution built on:

Advanced Rule Engine: Advanced Detection allows users to create customised rules to filter these false alarms. These rules can be based on various factors like object size, shape, motion patterns, and specific zones within the monitored area.
Reduced False Positives: Advanced filtering and logic capabilities within the rule engine will enable the system to differentiate between genuine threats and irrelevant events, minimising false alarms and saving operators valuable time.

This focus on advanced detection rules will unlock a new level of sophistication for Calipsa. By enabling operators to tailor the alarms to specific scenarios, Calipsa will remain at the forefront of intelligent CCTV analysis, empowering law enforcement and security professionals to extract the most valuable insights from their video data.

The Process:
Iterative Development and User-Centric Design

Version 1: Proof of Concept

Version 1 - Proof of Concept

This initial version (v1) focused on a Proof-of-Concept (POC) approach to gather user feedback and inform future development. The v1 design offered functionalities for:

Color detection rules
Object counting rules
Crowd forming rules
Loitering detection rules

User Feedback & Learnings:

Color Detection: Customer feedback indicated minimal use for color-based detection rules. This functionality might be removed or reworked in future iterations.
Terminology Confusion: Users expressed confusion regarding the difference between "count," "crowd forming," and "loitering" rules. Clearer definitions and functionalities are needed for these features.
Machine Learning Integration: The Machine Learning (ML) team initiated research on training models for more accurate detection based on user-defined rules and their expectations.

Version 2: Minimum Viable Product (MVP)

Version 2 - MVP

This version builds upon the user feedback and learnings gathered from the v1 Proof-of-Concept (POC) concept testing.

Context & Goals:

This is the first public release of the product.
Simplified Rule Creation: The v2 interface prioritizes a streamlined rule creation process with a reduced number of initial choices. This allows for easier setup and faster debugging if necessary.
The goal is to encourage user adoption and enable smooth debugging of detection performance.

Design Decisions:

Modal Setup: A modal window approach was chosen for rule creation to provide a focused environment for users. This replaces the accordion-style menu from v1, which was perceived as overwhelming.
Simplified Rule Creation: The number of initial choices within the modal window was reduced to streamline the setup process. This allows for easier setup and faster debugging if necessary.
Appearance Matching Detection: The "Color Detection" functionality was replaced with "Watchlist" in v2. This feature enables users to define rules based on the appearance of objects within the video feed, offering more flexibility than color-based detection alone.
Phased Introduction: Complex functionalities like advanced scheduling will be introduced incrementally as user trust and system performance stabilise.

Learnings from v2:

Users expressed difficulty setting up rules due to a perceived overload of options.
Lack of scheduling features made users hesitant to rely on the system.
Building user trust remains a key challenge in the initial stages.

Version 3: Streamlined Setup

Version 3 - Streamline setup

Goals for v3:

Streamline the rule setup process for improved user experience.
Introduce scheduling functionalities to enhance system flexibility and user trust.

Solutions Implemented:

3-Step Wizard Modal: The rule creation process is divided into a guided, three-step wizard modal. This reduces cognitive load and allows users to focus on one task at a time.
Integrated Scheduling: Scheduling functionality is introduced within the wizard modal, ensuring a smooth setup flow and enabling users to define rule activation times.
Rule Summary: A clear and concise summary of the defined rule is presented to users, setting clear expectations for the rule's outcome.

Outcomes:

User feedback indicates a higher adoption rate due to the simplified setup process.
The introduction of scheduling features is expected to further increase user trust and system utilisation.

Challenges Identified:

Despite improvements, user feedback suggests there is still "friction" for mass deployment. Further investigation is required to understand the underlying reasons.

Version 4: Bulk Rule Management

Version 4 - Flow to manage bulk enabling

This version builds upon the user feedback and learnings gathered from the v3 release, specifically addressing the "friction" for mass deployment.

Challenges in v3:

User feedback from v3 indicated remaining challenges for large-scale deployments.
Setting up rules for multiple cameras individually was perceived as time-consuming.

Solution Implemented:

Bulk Rule Creation: v4 introduces a workflow for creating rules and applying them to multiple cameras simultaneously. This streamlines the setup process for large deployments.
Integrated Scheduling: Scheduling functionality remains integrated within the bulk rule creation flow, allowing users to define activation times for their rules across multiple cameras.

Outcomes:

Bulk rule creation is expected to significantly reduce the time and effort required for mass deployment, paving the way for the Sales team to sell the product.
Combining bulk rule creation with scheduling offers a comprehensive solution for managing detection across multiple cameras.
The simplified deployment process empowers the sales team to present Advanced Detection as a more user-friendly and efficient solution, potentially increasing sales opportunities.

Future Plans

For version 5, the product will introduce the new detection capabilities. The ML team has started the training pipeline and the model should be ready for beta testing in 2025.

Conclusion

Through iterative design and close attention to user feedback, we transformed Calipsa's Advanced Detection Rules from a basic concept into a powerful, user-friendly tool for security professionals. Each version brought us closer to our goal of providing sophisticated, customizable CCTV analysis that adapts to the complex needs of modern security environments.

This project showcases the importance of:

Iterative development based on user feedback
Balancing advanced features with ease of use
Addressing specific user pain points (e.g., mass deployment)
Continuous improvement and forward-thinking (e.g., ML integration)

By maintaining this user-centric, iterative approach, we've positioned Calipsa to remain at the forefront of intelligent CCTV analysis, empowering security professionals to extract maximum