{
  "id": 486503,
  "title": "How Two-Stage Object Detectors Went From 47 Seconds to Real-Time-Adjacent",
  "url": "https://urgent.news/2026/08/10/how-two-stage-object-detectors-went-from-47-seconds-to-real-time",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-08-10T16:45:06.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/sakramen/how-two-stage-object-detectors-went-from-47-seconds-to-real-time-adjacent-4m2"
  },
  "original_language": "en",
  "account": null,
  "summary": "Object detection has two broad architectural families: one-stage detectors that predict boxes and classes directly in a single pass, and two-stage detectors that first propose candidate regions, then classify and refine them. Two-stage detectors were the dominant approach from 2014 to 2017 and are still the most accurate choice on many benchmarks when latency isn't the constraint. The interesting…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}