#人脸识别#PaddleDetection为基于飞桨PaddlePaddle的端到端目标检测套件,提供多种主流目标检测、实例分割、跟踪、关键点检测算法,配置化的网络模块组件、数据增强策略、损失函数等,推出多种服务器端和移动端工业级SOTA模型,并集成了模型压缩和跨平台高性能部署能力,帮助开发者更快更好完成端到端全开发流程。
#安卓#⚡️An Easy-to-use and Fast Deep Learning Model Deployment Toolkit for ☁️Cloud 📱Mobile and 📹Edge. Including Image, Video, Text and Audio 20+ main stream scenarios and 150+ SOTA models with end-to-end ...
🍅🍅🍅YOLOv5-Lite: Evolved from yolov5 and the size of model is only 900+kb (int8) and 1.7M (fp16). Reach 15 FPS on the Raspberry Pi 4B~
翻译 - shufflev2-yolov5:更轻、更快、更容易部署。由yolov5进化而来,模型大小只有1.7M(int8)和3.3M(fp16)。当输入大小为320×320时,它可以在树莓派4B上达到10+ FPS~
ONNX Model Exporter for PaddlePaddle
使用ONNXRuntime部署PicoDet目标检测,包含C++和Python两个版本的程序
PyTorch Implementation of Backbone of PicoDet
#计算机科学#PP-PicoDet for Jetson Nano
#计算机科学#PP-PicoDet for Raspberry Pi 4