Google's PlaNet 10 years ago: Image-GPS geolocation too costly for practical use
giffmana · x · 2026-08-23
Discussion on using Image-GPS contrastive learning from scratch. The author notes that Google derisked the technical feasibility a decade ago with PlaNet (achieving superhuman accuracy), but concluded it would not pay back its cost in terms of usefulness, viewing it as an expensive toy/demo rather than a scalable supervision solution.
More from Research
- Paper proposes method for Jensen-Shannon centroid of categorical distributions — FrnkNlsn · 2026-08-23
- Building a text watermark remover for Claude, rewriting works best — Imaginary_Dinner2710 · 2026-08-23
- RSI replaces scaling as core AI strategy in 2026; AIBuildAI emerges — 机器之心 · 2026-08-23
- PlayWorld: Benchmarking world models via long-horizon agent gameplay — 机器之心 · 2026-08-23
- MIT Professor on AI for Science: The age of abundant discovery is here — ProfBuehlerMIT · 2026-08-23
- Qwen3.8-27B Quantization Analysis: 4-bit Damage Peaks at Context Start — sadnessdevil · 2026-08-23