# Field from Above

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Tomasz Kneblewski](https://cerebralvalley.ai/u/tkneblewski)
- **GitHub:** https://github.com/kneblewskitomasz-sys/field-from-above
- **Demo video:** https://youtu.be/n3nTe37zQ1I
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/722

Field from Above turns a Polish parcel number into a dated, cloud-masked NDVI/NDRE map, an orthophoto with SAM 3.1 masks, and a short note that says what was measured, how uncertain it is, and where to walk first. I own five small fields, from 0.30 to 2.88 hectares. Satellite data covers them every few days, but a farmer cannot walk into a satellite image, and an agronomist is expensive for a 0.3-hectare strip. Every number carries its date, pixel count, cloud share and confidence interval. A separate fact check rejects any model text that adds a number or a date outside the measurement. When data is missing, the tool says so: parcel 90/1 returns a named absence, and a cloudy scene returns a cloud status instead of an old image shown as current. One date can lie, so the tool also reads the whole 2026 season and the national terrain model: a darker spot on parcel 92/10 looked like a problem on 26 October, but it came back in only 22 of 292 pixels (7.5%, threshold 10%), so the verdict is "does not return" and the note says not to start there. It is not a soil test and it does not recommend fertiliser. Live demo: https://kneblewskitomasz-sys.github.io/field-from-above/demo/

## More from Meta Global AI Developer Hackathon

- [MANISA](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/719)
- [Avasara Lab](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/720)
- [VANTALab](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/721)
- [EdgeCases](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/723)
- [RAJAKUMARAN](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/724)
- [Point, Say, Press](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/725)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/722. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt