emptysoal/cuda-image-preprocess
Speed up image preprocess with cuda when handle image or tensorrt inference
This project helps operations engineers, MLOps specialists, or anyone deploying computer vision models accelerate the crucial image preparation steps before feeding images into a neural network for inference. It takes raw images and quickly transforms them (e.g., resizing, color conversion, normalization) using NVIDIA CUDA, then feeds these optimized images into a TensorRT-powered inference engine. This process significantly speeds up the overall image analysis workflow.
Use this if you are deploying computer vision models with TensorRT and need to drastically reduce the time taken for image preprocessing steps, which often become a bottleneck.
Not ideal if you are working with non-image data, do not use NVIDIA GPUs or TensorRT for inference, or your current image preprocessing speed is already sufficient for your application.
Stars
84
Forks
7
Language
Cuda
License
MIT
Category
Last pushed
Nov 05, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/computer-vision/emptysoal/cuda-image-preprocess"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.