zalo/dataset-zero-obstacle-high-density-z01

This project generates practice problems and solutions for software that automatically routes electrical connections on printed circuit boards (PCBs). It creates random connection puzzles on a 10x10 millimeter square board, where connections need to be made between points on the edges without any obstacles in the way. For each puzzle, it saves the solution and an image showing both the required connections and the final routed paths. The code works by randomly generating a set of boundary connection pairs, then attempting to solve the routing problem using a built-in solver. When it successfully finds a solution, it saves the results into a data file and creates images showing the initial connections and the completed routing. If the solver cannot solve a specific puzzle, it simply skips that one and logs the failure. The output includes a structured data file with all the successful solutions, along with visual representations in both vector and pixel image formats. PCB designers and developers working on circuit board autorouting software would find this useful for training or testing their routing algorithms. For example, if someone is building AI tools to automatically lay out circuit board wiring, they need thousands of examples to train their system. This project provides those examples automatically, each one containing a specific routing challenge alongside its correct solution, along with visual diagrams showing what the board looks like before and after the paths are drawn. The project also includes a web service component that can be deployed to Cloudflare's edge network, allowing you to generate these datasets at scale across multiple concurrent requests. This is useful if you need to produce large volumes of training data quickly, since you can request thousands of samples from a remote endpoint instead of generating them all on your local machine. You can control how many samples to generate and how many requests to run in parallel.