Improve Python ALPR Script Accuracy
Budget: $250 – $750 USD
I’m working on a python program to recognize license plate.
I already have a working Python-based Automatic License Plate Recognition app, but its accuracy drops when I feed it traffic-camera footage shot in both daylight and at night. Plate numbers themselves are the main sticking point; state labels and special symbols aren’t critical right now.
Here’s what I need:
• Swap in—or bolt on—a more robust detection + OCR model that keeps plate-number accuracy high under mixed lighting. If you already own or have trained such a model, I’m happy to buy it rather than reinvent the wheel.
• Integrate the upgraded model into my existing script (OpenCV + standard Python libs) with minimal upheaval to the current codebase.
• Provide a clear implementation outline: dependencies, configuration steps, and a short test driver so I can benchmark daytime vs. nighttime clips. .
Deliverables are the updated script, the model weights (or install link), and a concise README. I’ll handle deployment once I can confirm improved precision on my mixed-lighting sample set.
What I’m looking for is exactly in the first 20 seconds of this video.
BONUS POINTS : if you have done a project like this and can show example
https://youtu.be/fyJB1t0o0ms?si=60zqZpK_mMmx8QUA
I already have a working Python-based Automatic License Plate Recognition app, but its accuracy drops when I feed it traffic-camera footage shot in both daylight and at night. Plate numbers themselves are the main sticking point; state labels and special symbols aren’t critical right now.
Here’s what I need:
• Swap in—or bolt on—a more robust detection + OCR model that keeps plate-number accuracy high under mixed lighting. If you already own or have trained such a model, I’m happy to buy it rather than reinvent the wheel.
• Integrate the upgraded model into my existing script (OpenCV + standard Python libs) with minimal upheaval to the current codebase.
• Provide a clear implementation outline: dependencies, configuration steps, and a short test driver so I can benchmark daytime vs. nighttime clips. .
Deliverables are the updated script, the model weights (or install link), and a concise README. I’ll handle deployment once I can confirm improved precision on my mixed-lighting sample set.
What I’m looking for is exactly in the first 20 seconds of this video.
BONUS POINTS : if you have done a project like this and can show example
https://youtu.be/fyJB1t0o0ms?si=60zqZpK_mMmx8QUA