FITSHIELD

Home / Downloads / Presentations & Slides

Presentation Slides & Oral Defense Decks

Official slide decks, presentation outlines, defense agendas, and technical viva slide walkthroughs across the FITSHIELD academic evaluation lifecycle.

6
Tracked Slide Decks
Available
Technical Viva Reference
16:9 HD
Widescreen Format
R26-IT-099
SLIIT Project ID
Milestone 1 • Proposal Defense 20 Min Defense + 10 Min Q&A

Project Proposal Presentation

Problem Space, Literature Survey Gap & 4-Component Intelligence Architecture

Defense Date
16 - 18 March 2026
Slide Count
32 Slides (16:9)
Presenters
All 4 Group Members
Evaluation Panel
Faculty Assessment Board

Key Presentation Agenda:

  • • Background & Problem: Critical health hazards in unsupervised fitness regimens, drug-nutrient conflicts, and habit abandonment.
  • • Literature Survey Gap: Limitations of generic fitness trackers lacking pharmacological integration and vision biomechanics.
  • • 4-Component Architecture: Diet Safety, Supplement Advisory, Posture Vision, and Adaptive Habit Evolution (AHEE).
  • • Work Breakdown & Budget: Milestone roadmap, compute resource allocations, and ethical compliance sign-offs.
Scheduled (March 2026)
Milestone 2 • Midterm Review 1 20 Min Review + Q&A

Progress Presentation 1

System Design, Curated Datasets & Component Prototype Benchmarks

Review Date
11 - 13 May 2026
Slide Count
28 Slides (16:9)
Presenters
All 4 Group Members
Target Progress
40% Complete Prototypes

Key Presentation Agenda:

  • • Dataset Integrity: Pharmacological CYP450 enzyme interaction tables & 3,700+ supplement sample collection.
  • • Computer Vision Skeleton Baseline: MediaPipe BlazePose landmark extraction pipeline running on web webcam streams.
  • • Early Model Training: Random Forest supplement classifier feature importance selection & hyperparameter setup.
  • • AHEE Habit Modeling: Markov state definition for workout adherence recalibration loops.
Scheduled (May 2026)
Milestone 3 • Midterm Review 2 25 Min Demo + Evaluation

Progress Presentation 2

90% System Integration, Real-Time Vision Telemetry & Cross-Validation

Review Date
31 Aug - 02 Sep 2026
Slide Count
36 Slides (16:9)
Presenters
All 4 Group Members
Target Progress
90% Integrated Functionality

Key Presentation Agenda:

  • • Cross-Component Orchestration: Unified REST/JSON API gateway communicating across all 4 microservices.
  • • Live Vision Demo: Real-time joint angle trigonometry (knee flexion, sagittal spine) at 30+ FPS with immediate audio cues.
  • • Model Accuracy Metrics: Random Forest 93.4% cross-validated accuracy and confusion matrix analysis.
  • • Safety Hardening: Automatic blocker test cases ensuring unsafe supplement/diet recommendations are rejected.
Scheduled (August 2026)
Milestone 4 • Final Thesis Defense 30 Min Defense + Panel Viva

Final Assessment & Oral Defense

End-to-End System Demonstration, Empirical Validation & Academic Findings

Defense Date
14 October 2026
Slide Count
45 Slides (16:9)
Presenters
All 4 Members & Supervisors
Evaluation Panel
External & Internal Examiners

Key Presentation Agenda:

  • • Final System Demonstration: Complete patient-athlete onboarding, meal logging, supplement optimization, and live workout telemetry.
  • • Empirical Results & Benchmarks: Precision, recall, inference latency, frame-drop tolerance, and behavioral adherence lift.
  • • Research Contributions: Formal novelty contribution to dual-layer safety fitness computing.
  • • Commercialization & Future Scope: Wearable ECG sensor sync, LLM conversational querying, and multi-gym deployment.
Scheduled (October 2026)
Available • Technical Viva Notes Official Viva Architecture

Supplement Advisory Backend & ML Model

Random Forest Classification (93.4%), REST Endpoints & Contraindication Pipeline

Component Lead
W.M.P.G.S. Dilhara
Student ID
IT22169044
Document Format
PDF Technical Reference
Availability
Verified & Ready

Key Presentation Agenda:

  • • System Request Flow: High-performance architecture showing client requests routed to the inference engine.
  • • Machine Learning Pipeline: Random Forest Classifier trained across 3,788 multi-attribute records with 93.4% accuracy.
  • • REST API Contracts: POST /api/v1/recommend-supplement and POST /api/v1/check-contraindication.
  • • Pharmacological Safety Layer: Automatic contraindication checking against active cardiovascular and stimulant limits.
Download PDF Reference
Technical Viva Reference 15 Min Code Walkthrough

Posture Recognition & Form Correction

MediaPipe BlazePose 33-Landmark Kinematics & Real-Time Audio Correction

Component Lead
H.M.T.N. Herath
Student ID
IT22248176
Slide Count
26 Slides (16:9)
Viva Defense
19 October 2026

Key Presentation Agenda:

  • • Edge Vision Inference: BlazePose pipeline optimizing landmark stability under varied lighting and occlusions.
  • • Trigonometric Joint Calculation: Real-time vector angle formulas for knee flexion, hip hinge, and lumbar spine alignment.
  • • Real-Time Audio Coaching: Synthesizing immediate auditory correction feedback within 120ms of form deviation.
  • • Repetition Counting State Machine: Peak-inflection hysteresis filtering eliminating false repetition counts.
Prepared for Viva

Oral Defense Structure & Faculty Assessment Rubrics

FITSHIELD research milestone presentations follow the official SLIIT Faculty of Computing examination protocols:

Time Distribution

Each milestone allocates 20 minutes for core slides, 10 minutes for live operational demonstrations, followed by 15 minutes of panel cross-examination.

Individual Accountability

Every group member independently defends their dedicated component's technical implementation, mathematical model, and empirical validation metrics during the oral viva.

Live Demonstration

Presentations require verified live executions of computer vision skeleton tracking, drug-nutrition interaction alerts, and the Random Forest inference API.

Academic Rigor

Slide decks cite published biomedical datasets (DrugBank, PubMed, BioMed) and computer vision benchmarks (COCO, MPII Human Pose) supporting the design.

Cite FITSHIELD Research Presentations

Use this reference format when citing FITSHIELD oral defenses and technical presentations:

@misc{fitshield_presentation_2026,
  title        = {FITSHIELD: Multi-Agent Intelligence for Medication-Aware Nutrition, Posture Vision, and Adaptive Habit Formation [Oral Presentation]},
  author       = {Ramanayake, R.M.S.M. and Dilhara, W.M.P.G.S. and Herath, H.M.T.N. and Gamaachchige, C.D.},
  howpublished = {Faculty of Computing, Sri Lanka Institute of Information Technology (SLIIT)},
  year         = {2026},
  note         = {Final Year Research Project ID: R26-IT-099}
}