SUMIP COMPOSITES

Pvt. Ltd. | Ahmedabad, Gujarat

FluxAI 360° Vision Inspection

AI-Powered Quality Inspection System

January 2026

Confidential

Prepared by

Trinesis Technologies Pvt. Ltd.

01 Executive Summary

We propose to implement a FluxAI 360° Vision Inspection System for Sumip Composites' GRP/FRP pultrusion lines. This AI-powered system will replace manual sampling with 100% automated inspection, capturing all surfaces and generating tamper-proof digital quality certificates.

Project Scope

Complete AI-powered inspection system with 6-camera 360° coverage, real-time defect detection, and digital certification.

Products Covered

C-Channels and Flat Plates - the primary product types in your pultrusion lines.

Key Innovation

360° inspection from all sides, AI defect detection with >90% accuracy, and cryptographically signed quality certificates.

Deployment Options

Flexible installation - Online (per production line) or Offline (central QC hub for multiple lines).

02 The Challenge

You operate 16 pultrusion lines running 24x7. Today, quality inspection is manual and sample-based.

⚠ Current Problems

  • Operators inspect by eye - Quality varies by shift and person
  • Only samples are checked - Defects slip through to customers
  • No documentation - Cannot trace issues or prove quality
  • Glossy surfaces - Hard to see defects under factory lighting
  • Heavy dust environment - Affects visibility and consistency

✓ What FluxAI Delivers

  • 360° inspection - 6 cameras capture all surfaces
  • 100% coverage - Every product inspected, not just samples
  • Quality certificates - Digitally signed, tamper-proof, verifiable online
  • Polarized lighting - Handles glossy surfaces without glare
  • IP65 enclosure - Dust-proof, factory-ready hardware

03 Our Solution

FluxAI inspects every product from all sides, catches defects automatically, and generates digital quality certificates your customers can verify.

360° Camera Layout

📷 6-Camera 360° Coverage

  • CAM 1: Top surface
  • CAM 2: Bottom surface
  • CAM 3: Left side / web outside
  • CAM 4: Right upper (45° angle)
  • CAM 5: Right center / web inside
  • CAM 6: Right lower (45° angle)

🤖 AI-Powered Detection

  • Deep learning models trained on your defects
  • >90% detection accuracy on critical defects
  • <5% false positive rate
  • <100ms inference time per image
  • Configurable sensitivity thresholds

📄 Digital Certificates

  • Per-product or per-batch certificates
  • RSA-2048 cryptographic signature
  • QR code for instant verification
  • Embedded inspection images
  • Online verification portal

💻 Operator Dashboard

  • 15.6" industrial touch display
  • Real-time PASS/FAIL indication
  • Live camera feeds
  • Defect highlighting with bounding boxes
  • Shift reports and statistics

📊 Custom Reports

  • Custom reports tailored to your business needs
  • Shift-wise, line-wise, and defect-wise analytics
  • Export in PDF, CSV, and JSON formats
  • Scheduled reports via email
  • Flexible report builder for new requirements

How It Works

Inspection Box Detail
How It Works Flow
1
Product Enters

Product passes through the inspection box on conveyor or manual feed

2
Cameras Capture

6 cameras simultaneously capture all surfaces with polarized lighting

3
AI Analyzes

Deep learning models analyze images and detect defects

4
Certificate Generated

Digital certificate created with images, results, and cryptographic signature

Operator sees the result on screen. If FAIL, an alert sounds and they remove the product. All data is logged automatically.

04 Defects Detected

FluxAI detects the following defect types, categorized by severity:

1
Cracks / Fractures Auto-reject >1mm
2
Dry Fiber Exposure Auto-reject >1mm
3
Burns / Discoloration Auto-reject >1mm
4
Staple Inclusions Auto-reject any
5
Air Bubbles / Voids Flag for review >2mm
6
Delamination Flag for review >2mm
7
Surface Scratches Log only >3mm
8
Resin-Rich Areas Log only >3mm
9
Edge Defects Log only >3mm
Critical: Immediate FAIL, auto-reject
Major: Flagged for operator review
Minor: Logged for analytics

* Thresholds are configurable based on your quality standards

05 Project Timeline

Project Timeline

★ Pilot Success Criteria

>90%
Detection Accuracy
<5%
False Positives
>99%
System Uptime
Easy to Use

If we don't meet these targets, we retrain the model free until we do.

06 Why Trinesis

📍 We Visited Your Factory

We saw your lines, your products, your challenges before writing this proposal. This isn't a generic solution - it's designed for Sumip.

🎯 Built for Your Needs

360° coverage, glossy surface handling, digital certificates - every feature addresses your specific requirements.

📊 Pilot First Approach

Prove it works before you invest in full rollout. Start with one line or hub, validate results, then scale.

💪 We Guarantee Results

>90% accuracy or we retrain free. We work until pilot criteria are met, at no additional cost.

Our Commitment

Detection accuracy >90%
System uptime >99%
Same-day response for critical issues
Work until pilot criteria met

07 Next Steps

Next Steps
1
Choose

Capacity tier and deployment (Online/Offline)

2
Identify

Which line or area for pilot installation

3
Provide

Defect samples for AI model training

4
Order

Issue PO (pricing valid 30 days)

We'll take it from there.

Let's build quality you can prove.

08 Terms & Conditions

🔒 Confidentiality Agreement

This document and all accompanying materials (technical specifications, cost breakdowns, certificates, UI mockups, and drawings) are the confidential and proprietary property of Trinesis Technologies Pvt. Ltd.

This proposal is shared exclusively with Sumip Composites Pvt. Ltd. for the sole purpose of evaluating the FluxAI 360° Vision Inspection System. The recipient agrees to:

  • Not disclose, distribute, or share any part of this document with third parties without prior written consent from Trinesis Technologies.
  • Use the information contained herein solely for the purpose of evaluating and implementing the proposed solution.
  • Not reverse-engineer, replicate, or use the technical designs, AI methodologies, or proprietary processes described in this proposal.
  • Return or destroy all copies of this document upon request or if the proposal is not accepted.

Any unauthorized use, reproduction, or distribution of this document may result in legal action under applicable Indian intellectual property and confidentiality laws.

Document Reference: TRIN/PROP/SUMIP/2026-01  |  Classification: Confidential  |  Valid Until: 30 days from date of issue

🔒

Trinesis Technologies

Confidential Proposal