DeepvizorRequest a demo

[ DEEPVIZOR JOURNAL ]

Video analytics.
A closer look.

Research, algorithms and practice. Explore the technology behind an intelligent camera.

10

publications
From ideas to deployment

All publications

06—10 / 10
  1. CECutMix76.876.9ID accuracy ↑60.565.4AURC ID ↓71.577.1AUROC OOD ↑56.553.9DS-F1 ↑32.539.8DS-AURC ↓
    06 / RESEARCH / ARXIV 2026

    SURE+: joint reliability assessment for classification

    From misclassifications to outliers: two reliability scores, four training components and SURE+ experimental results.

    · 7 min read · Read article
  2. Original research results
    07 / RESEARCH / ICLR 2026

    FSOD-VFM: few-shot object detection with graph diffusion

    Vision foundation models and graph diffusion detect whole objects from a few examples without fine-tuning. Method and ICLR 2026 results.

    · 7 min read · Read article
  3. Open metal cabinet containing a compact compute device and network equipment in a factoryIllustration / Deepvizor
    08 / ON-SITE ANALYTICS

    On-site video analytics: processing video locally

    How compact models support on-site video analysis, and what to check when choosing a compute device.

    · 2 min read · Read article
  4. Workshop frame with a worker bending to pick up a cartonIllustration / Deepvizor
    09 / MOTION ANALYSIS

    Human action recognition in video: how AI works

    One frame shows a pose. A sequence reveals an action. Why motion and context matter for video analytics.

    · 2 min read · Read article
  5. Warehouse yard frame with workers in hi-vis vests and a delivery vanIllustration / Deepvizor
    10 / COMPUTER VISION

    AI video analytics: how cameras recognize objects

    From pixels to people, vehicles and events. What happens behind a detection box, and how to test the result at your site.

    · 2 min read · Read article
Have a video analytics project in mind?info@deepvizor.com ↗