Lakhasly

3.7 Conclusion of Chapter 3 This chapter has presented a comprehensive analysis of the problem domain, the methodological framework, and the system requirements that together form the foundation for the design and implementation of the Intelligent Waste Sorting Assistant.Non functional requirements - usability, reliability, maintainability, performance, and design constraints - were specified in detail, mandating a client side web architecture using JavaScript, TensorFlow.js, and Supabase, with no image uploads and offline classification capability.The analysis began with a Fishbone diagram, which systematically decomposed the multifaceted causes of inefficient waste sorting into four thematic categories: People, Methods, Materials/Equipment, and Environment.Quantitative data on sorting accuracy will be gathered through pre and post intervention measurements, while qualitative insights from interviews, observation, and focus groups informed user centered design decisions.

3.7 Conclusion of Chapter 3
This chapter has presented a comprehensive analysis of the problem domain, the methodological framework, and the system requirements that together form the foundation for the design and implementation of the Intelligent Waste Sorting Assistant. The analysis began with a Fishbone diagram, which systematically decomposed the multifaceted causes of inefficient waste sorting into four thematic categories: People, Methods, Materials/Equipment, and Environment. This exercise revealed that the most critical and actionable gap is the absence of an immediate, accurate, and accessible guidance system at the point of disposal – a gap that a client side AI application is uniquely positioned to fill.
The research methodology adopted a mixed methods approach, combining quasi experimental and exploratory strategies. Quantitative data on sorting accuracy will be gathered through pre and post intervention measurements, while qualitative insights from interviews, observation, and focus groups informed user centered design decisions. The data collection procedure was structured in four phases: literature analysis, dataset preparation (using TrashNet and TACO with augmentation), requirement elicitation, and system evaluation. The Agile Scrum methodology was selected for development, recognizing the inherent uncertainty of AI model performance and the need for iterative refinement.
The system requirements were formalized in an IEEE 830 1998 compliant Software Requirements Specification, emphasizing privacy as a non negotiable attribute. Key functional requirements include real time image capture and processing (FR1), waste classification with a confidence threshold (FR2), disposal instruction display (FR3), and educational content (FR4). Non functional requirements – usability, reliability, maintainability, performance, and design constraints – were specified in detail, mandating a client side web architecture using JavaScript, TensorFlow.js, and Supabase, with no image uploads and offline classification capability.
The requirements model centered on a use case diagram with the User as the sole actor and five use cases, the most critical being “Classify Waste Item Using Camera”. A detailed narrative description of this use case specifies the preconditions, postconditions, basic and alternative flows, exception paths, and performance targets. Hardware and software specifications were defined, emphasizing browser compatibility and the absence of platform specific dependencies.
The development tools were selected to align with the privacy first, offline capable architecture. The programming language stack consists of HTML5, CSS3, and JavaScript (ES6), with TensorFlow.js for on device AI inference. The database strategy employs a dual layer system: localStorage for temporary, offline history (limited to ten records) and Supabase for persistent cloud storage of text metadata only. Additional tools include Git for version control, Vite for local development, and static hosting for deployment.
Finally, the narrative of the analysis method detailed the iterative journey from problem decomposition to actionable specifications. Key challenges – resolving the tension between educational content and interface simplicity, pivoting from a server based to a purely client side architecture, establishing a 70% confidence threshold, designing a keyword based mapping engine with stemming, and implementing a hybrid storage solution – were all addressed through systematic investigation and prototyping. The outcome is a clear, feasible project scope that directly informs the design presented in Chapter 4 and the implementation described in Chapter 5. Through this rigorous analysis, the project is firmly grounded in both user needs and technical realities, ensuring that the subsequent development phases are purpose driven and evidence based.