Dziennik zmian

Śledź najnowsze aktualizacje, nowe funkcje i ulepszenia VerifyReal.

AlphaObecnie w fazie testów alpha
5

v0.5.1

28 stycznia 2026

Engine Versioning & Reprocessing

  • 🆕 NoweEngine Versioning - Track which detection engine version was used for each analysis
  • 🆕 NoweAdmin Reprocess - Re-analyze images with the latest engine when updates are deployed
  • 🆕 NoweVersion Filtering - Filter analyses by engine version in admin panel
  • ✨ UlepszoneAdmin API now shows engine version and reprocess history for each analysis
5

v0.5.0

28 stycznia 2026

Shareable Results

  • 🆕 NoweShare Result - Share your analysis on social media with a beautiful preview card
  • 🆕 NoweShare Card Generator - 1200x630 PNG optimized for Facebook, LinkedIn, X, and Reddit
  • 🆕 NoweShare Page - Dynamic OG meta tags for proper social media previews
  • ✨ UlepszoneVLM inference now 47% faster with vLLM as primary tier 1
4

v0.4.0

27 stycznia 2026

Face Deepfake Detection

  • 🆕 NoweFace Deepfake Detector - Specialized model for detecting face manipulations
  • 🆕 NoweML Worker integration in forensic pipeline
  • ✨ UlepszonevLLM as primary labeling server for faster dataset processing
3

v0.3.0

20 stycznia 2026

Hybrid Taxonomy v2.0

  • 🆕 Nowev2.0 Hybrid Classification - Primary Subject + Scene Context taxonomy
  • 🆕 Nowe9 subject categories: Person, Animal, Vehicle, Food, Product, Text, Scenery, Art, Other
  • 🆕 Nowe3 scene contexts: Indoor, Outdoor, Studio
  • 🐛 PoprawkaFixed domain classification confusion between aspect ratio and content type
  • ✨ UlepszoneBetter forensic weights per domain for more accurate detection
2

v0.2.0

10 stycznia 2026

Multi-Layer Forensics

  • 🆕 Nowe10-layer forensic analysis pipeline
  • 🆕 NoweVLM-first tiered architecture with M4 Mac Mini cluster
  • 🆕 NoweAnomaly detection with AI reasoning
  • 🆕 NoweVisual heatmaps for ELA, frequency, and attention
  • ✨ UlepszoneThree-way classification: authentic / manipulated / ai_generated
1

v0.1.0

15 grudnia 2025

Initial Alpha Release

  • 🆕 NoweCore detection engine with ONNX ML ensemble
  • 🆕 NoweMetadata/EXIF analysis layer
  • 🆕 NoweError Level Analysis (ELA)
  • 🆕 NoweFrequency domain analysis (FFT)
  • 🆕 NoweMulti-language support: EN, NL, DE, FR, RO
  • 🆕 NoweUser feedback system for model improvement

Co dalej?

Sprawdź naszą mapę drogową, aby poznać nadchodzące funkcje, takie jak wykrywanie wideo, rozszerzenia przeglądarki i więcej.

Zobacz Roadmap

Chcesz zgłosić błąd lub zasugerować funkcję? Skontaktuj się z nami