PlasticVision — Industrial AI Quality Inspection Platform logo
sensifai

PlasticVision — Industrial AI Quality Inspection Platform

Grounded, auditable AI visual inspection for plastics injection molding and extrusion. SmartMold QA defect detection, shadow-mode verification, and SPC analytics.

O tym modelu

PlasticVision: Industrial AI Inspection for Plastics Manufacturing

PlasticVision empowers injection-molding and extrusion manufacturers to turn existing machine cameras into grounded, auditable, and advisory process intelligence. Built specifically for high-throughput plastics production, PlasticVision automates visual defect detection, short-shot and flash localization, un-ejected part verification, and mold protection alerts without requiring on-site vision engineers or custom machine-level coding.


Key Capabilities & Architecture

1. SmartMold QA Cavity Inspection

  • Automated Mold Monitoring: Optical inspection of mold cavities and ejector pins between machine cycles to verify full part clearance.
  • Defect Classification: Detects flash, short shots, sink marks, flow lines, jetting, burn marks, and surface blisters in real time.

2. Rigorous Shadow-Mode Onboarding

  • Empirical Validation Gate: Operates silently in shadow mode on historical or live cycles to measure exact escape and false-positive rates before going live.
  • Baseline Tensor Training: Establishes golden baselines from 20–50 verified good cycles to minimize false alarms under ambient factory lighting.

3. Safety Boundary by Design

  • Strict Advisory-Only Architecture: All alerts and metrics carry safety_classification: advisory. Never arrests safety-rated motion and refuses safety-interlock integrations (EUROMAP 67/81).
  • Non-Intrusive Integration: Communicates via standard industrial telemetry protocols (EUROMAP 77, 82, 84, OPC UA, and MQTT) to advise operators and non-safety pneumatic reject gates.

4. Grounded AI & Knowledge Base Retrieval

  • Anti-Hallucination Invariants: Powered by centralized Gemini multimodal vision models with versioned JSON Schema validation.
  • Root-Cause Hypotheses: Correlates visual defect signatures with polymer melt temperature, injection pressure, and cycle time drift to provide actionable troubleshooting advice.

5. Native Sensifai Ecosystem Integration

  • Logto OIDC Single Sign-On: Secure enterprise authentication with PKCE authorization code flow.
  • Gatekeeper Metered Billing: Transparent per-cycle usage tracking integrated with customer prepaid wallets.

Target Manufacturing Applications

  • Automotive Plastics: Interior trim, under-the-hood housings, connectors, and precision engineered resin components.
  • Medical Plastics: Sterile syringe barrels, pipette tips, fluid connectors, and diagnostic cassettes requiring pristine optical quality.
  • Consumer & Industrial Packaging: Caps, closures, thin-wall food containers, and cosmetic packaging.
  • Technical Injection Molders: High-cavitation tooling where manual end-of-shift inspection misses intermittent cavity defects.

Kluczowe funkcje

SmartMold QA real-time cavity visual inspection and un-ejected part detection
Shadow-mode validation pipeline proving escape and false-positive rates before go-live
Strict advisory-only architecture with zero safety-interlock violations
Grounded AI defect classification for short shots, flash, sink marks, and burns
Real-time SPC quality analytics and shift-level OEE yield tracking
Multi-protocol telemetry ingestion supporting EUROMAP 77/82/84 and OPC UA

Przypadki użycia

  • Injection molding mold protection and un-ejected part detection
  • Automotive plastic component surface defect and flash inspection
  • Medical consumable high-cavitation tooling quality verification
  • Packaging closure and cap visual inspection at line speeds

Szybkie informacje

Dostawcasensifai
Dostęp do APIREST API

Potrzebujesz pomocy?

Masz pytania dotyczące tego produktu? Nasz zespół chętnie pomoże.

Skontaktuj się z nami

Plany cenowe

Plany cenowe

marketplace.flexible

Cavity Cycle Inspection Metered

marketplace.predictableBilling

Płatność za użycie
0.003/Request