Do you any plan to improve HOG based object detection by using CUDA?
贡献者指南
技术栈
cpp
领域
machine learning
议题类型
feature
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
5
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
over 1 week
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
stale
清晰度议题是否清楚说明期望改动、验收标准和下一步。
clear
前置要求
C++ experienceCUDA knowledgeObject detection basics
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
10
研究方向
Review the existing HOG implementation in dlib (e.g., dlib/image processing/hog.h) to understand current performance. Investigate potential CUDA integration points by examining other CUDA modules in dlib (if any). Consider creating a benchmark to measure current HOG performance and propose a design for CUDA acceleration. Engage with maintainers in the issue comments to clarify requirements and feasibility.
