Thanks for the thoughtful feedback! I agree that some of these drawbacks aren't inherent limitations of OpenCV or C++. The goal isn't to argue that OpenCV is bad, but to explore whether we can eliminate heavyweight dependencies like OpenCV and ONNX Runtime entirely for our OCR use case.
By rebuilding the image processing and inference pipeline in pure Rust, we aim to reduce binary size, simplify cross-platform builds, and make deployment across server, WebAssembly, and mobile more straightforward.
Get Muhamad Rizki’s stories in your inbox
Join Medium for free to get updates from this writer.
I agree that OpenCV can be optimized and that Rust doesn't automatically eliminate memory or concurrency issues. Ultimately, this is about reducing dependency overhead while maintaining competitive performance, not claiming that Rust is universally superior to C++.
