Live data from Hacker News

Show HN: Satellite imagery object detection using text prompts

useful-ai-tools.com

1–10 of 27 posts

Show HN: Satellite imagery object detection using text prompts

#1
I built a browser-based tool for detecting objects in satellite imagery using vision-language models (VLMs). You draw a polygon on the map and enter a text prompt such as "swimming pools", "oil tanks", or "buses". The system scans the selected area tile-by-tile and returns detections projected back onto the map as GeoJSON.

Pipeline: select area and zoom level, split the region into mercantile tiles, run each tile with the prompt through a VLM, convert predicted bounding boxes to geographic coordinates (WGS84), and render the results back on the map.

It works reasonably well for distinct structures in a zero-shot setting. occluded objects are still better handled by specialized detectors like YOLO models.

There is a public demo and no login required. I am mainly interested in feedback on detection quality, performance tradeoffs between VLMs and specialized detectors, and potential real-world use cases.

Show HN: Satellite imagery object detection using text prompts
useful-ai-tools.com

Re: Show HN: Satellite imagery object detection using text prompts

#6
very cool

Once I figured out how to use the UI I did 2 scans. first one I had to zoom in before the identification boxes popped up. At first I thought it didnt do anything

Second scan I put over a local aviation museum with a mix of helicopters, unusual planes, cars, buildings, and other equipment. I was surprised to see everything identified correctly, though it missed a single helicopter.

I'd love a little bell or notification when the scan completes, as I hit 'scan', switch to a different tab and then forgot I was waiting

Re: Show HN: Satellite imagery object detection using text prompts

#7
post #5

Tangent question, I know of services like Planet Labs, Maxar... is the capability there now assuming you had money, where you could tag a ship from space and watch it travel (I know there is something like ADSB for ships) but would be interesting.

It's still at the "technical challenge to fuse data from AIS [ADSB for ships, complete with spoofing and transponders being turned off] to [generally] static images from multiple different sources" stage, at least unless you've got the budget for a video satellite to stick in a geostationary-ish orbit

Re: Show HN: Satellite imagery object detection using text prompts

#8
post #5

Tangent question, I know of services like Planet Labs, Maxar... is the capability there now assuming you had money, where you could tag a ship from space and watch it travel (I know there is something like ADSB for ships) but would be interesting.

It's still at the "technical challenge to fuse data from AIS [ADSB for ships, complete with spoofing and transponders being turned off] to [generally] static images from multiple different sources" stage, at least unless you've got the budget for a video satellite to stick in a geostationary-ish orbit

I also wasn't sure how good satellite image quality is where you can match a ship from above, maybe the pattern of the freight containers could help.

And track I didn't indicate frequency, it's not per minute, but say hourly.

Re: Show HN: Satellite imagery object detection using text prompts

#9
post #6

very cool Once I figured out how to use the UI I did 2 scans. first one I had to zoom in before the identification boxes popped up. At first I thought it didnt do anything Second scan I put over a local aviation museum with a mix of helicopters, unusual planes, cars, buildings, and other equipment. I was surprised to see everything identified correctly, though it missed a single helicopter. I'd love a little bell or…

Thanks for trying it out. The detections not all appearing before zooming is because I added a LOD (level of detail) rendering method, so if hundreds of thousands of objects are detected, it won’t crash the system. Only the areas you’re looking at render, and the more you zoom in, the more objects are displayed. It was a pain to set up, but it’s worth it. The notification idea is great, and I’ll add a sound to play when a scan finishes.

Re: Show HN: Satellite imagery object detection using text prompts

#10
post #6

very cool Once I figured out how to use the UI I did 2 scans. first one I had to zoom in before the identification boxes popped up. At first I thought it didnt do anything Second scan I put over a local aviation museum with a mix of helicopters, unusual planes, cars, buildings, and other equipment. I was surprised to see everything identified correctly, though it missed a single helicopter. I'd love a little bell or…

Thanks for trying it out. The detections not all appearing before zooming is because I added a LOD (level of detail) rendering method, so if hundreds of thousands of objects are detected, it won’t crash the system. Only the areas you’re looking at render, and the more you zoom in, the more objects are displayed. It was a pain to set up, but it’s worth it. The notification idea is great, and I’ll add a sound to play w…

Thanks, makes sense to me. I was just confused at first since i thought maybe it didnt do anything or my adblocker was making it go weird
Post reply on HN