Open-source analysts have long been stuck with an impossible choice: free satellite data that is too stale to track moving targets, or commercial imagery that costs more than most independent researchers can justify. This project cuts through that trade-off with a method that is almost elegant in its simplicity. By exploiting the one-second delay between Sentinel-2's visible bands, Drish turns a limitation into a feature. A truck moving at highway speed shifts roughly 22 meters between captures, leaving a telltale blue-green-red smear across a handful of pixels. The tool finds those smears, counts them, estimates speed and heading, and builds monthly volume trends. That is not a hack. That is physics, applied with intention.
What makes this worth your attention is not the novelty of the trick, though the trick is clever. It is the accessibility of the entire package. The tool runs locally as a FastAPI app with a browser dashboard, and the underlying detection model comes from peer-reviewed research published in *Remote Sensing of Environment* by Fisser et al. in 2022. That matters. It means the method is not a black box or a proprietary algorithm guarded behind a paywall. It is reproducible, auditable, and built on science that has already survived scrutiny. For analysts who need to trust their sources, that is a significant step up from relying on commercial providers who will not share their methodology or their error rates.
The practical implications extend well beyond military monitoring, which is where the original focus lies. The same technique can be applied to supply chain bottlenecks, port congestion, or even traffic flow on rural highways where ground sensors are sparse. The data is free, the compute is local, and the barrier to entry is a working knowledge of Python and satellite imagery. That lowers the floor for what constitutes credible open-source analysis. It also raises the ceiling on what independent researchers can contribute to conversations that have historically been dominated by governments and well-funded corporations.
Our take is straightforward: this is the kind of work that moves the field forward, not because it introduces brand-new technology, but because it repurposes existing tools in a way that expands who gets to participate. The author deserves credit for sharing the code, the model, and the documentation. That kind of openness is what turns a clever proof of concept into a usable resource. If you have been hesitating to explore satellite-based logistics tracking because of cost or complexity, this project removes both excuses. The GitHub link is in the post. The next step is to clone the repo and see what you can see.