使用ONNX+TensorRT部署人脸检测和关键点250fps

本文介绍 使用ONNX+TensorRT部署人脸检测和关键点250fps 使用ONNX+TensorRT部署人脸检测和关键点250fps This article was original written by Jin Tian, welcome re-post, first come with https://jinfagang.github.io . but please keep this copyright info, thanks, any question could be asked via wechat: jintianiloveu 我尽量用尽可能短的语言将本文的核心内容浓缩到文章的标题中,前段时间给大家讲解Jetson……

阅读全文

ONNXRuntime采用TensorRT后端快速推理

本文介绍 ONNXRuntime采用TensorRT后端快速推理 ONNXRuntime采用TensorRT后端快速推理 This article was original written by Jin Tian, welcome re-post, first come with https://jinfagang.github.io . but please keep this copyright info, thanks, any question could be asked via wechat: jintianiloveu onnxruntime是一个专门推理onnx模型的前向推理框架,它有一个好处就是模型统一、接口统一但是加……

阅读全文

ONNX里面的元运算

本文介绍 ONNX里面的元运算 ONNX里面的元运算 This article was original written by Jin Tian, welcome re-post, first come with https://jinfagang.github.io . but please keep this copyright info, thanks, any question could be asked via wechat: jintianiloveu 其实这个有点标题党,但这篇文章要做的事情,还是很有意义的,我们很多时候再用onnx转模型的时候,遇到的都不是一个层的问题,而往往是某个node不对,要么缺少attribut……

阅读全文

maskrcnn-benchmark转onnx再转TensorRT实录

本文介绍 maskrcnn-benchmark转onnx再转TensorRT实录 maskrcnn-benchmark转onnx再转TensorRT实录 This article was original written by Jin Tian, welcome re-post, first come with https://jinfagang.github.io . but please keep this copyright info, thanks, any question could be asked via wechat: jintianiloveu 这篇文章将记录一下,如何通过onnx将maskrcnn模型转到tenso……

阅读全文

pytorch里面的torch.gather操作

本文介绍 pytorch里面的torch.gather操作 pytorch里面的torch.gather操作 This article was original written by Jin Tian, welcome re-post, first come with https://jinfagang.github.io . but please keep this copyright info, thanks, any question could be asked via wechat: jintianiloveu torch.gather 只是一个引子,别看它简单,但能引出很多问题。我们先来看看,它是如何工作的。假如我们有一个矩阵: [[34, 4, 6], [45, 6, 7]] 我们想要……

阅读全文

FOTS和ROIRotate以及仿射变换

本文介绍 FOTS和ROIRotate以及仿射变换 FOTS和ROIRotate以及仿射变换 This article was original written by Jin Tian, welcome re-post, first come with https://jinfagang.github.io . but please keep this copyright info, thanks, any question could be asked via wechat: jintianiloveu FOTS是一个速度比较快,同时是一个端到端的文本检测方法,这个方法有一些东西值得学习,这也是写这篇文章的原因。通常我们实现一个文本检……

阅读全文

onnx-tensorrt实现添加自己的模型plugin

本文介绍 onnx-tensorrt实现添加自己的模型plugin onnx-tensorrt实现添加自己的模型plugin This article was original written by Jin Tian, welcome re-post, first come with https://jinfagang.github.io . but please keep this copyright info, thanks, any question could be asked via wechat: jintianiloveu 总结一下现在基本上深度学习模型部署的套路: 一般都是把模型training然后export 到onnx;……

阅读全文

本地编译Apollo Cyber系统

本文介绍 本地编译Apollo Cyber系统 本地编译Apollo Cyber系统 This article was original written by Jin Tian, welcome re-post, first come with https://jinfagang.github.io . but please keep this copyright info, thanks, any question could be asked via wechat: jintianiloveu 我们在玩Apollo的时候,其实大多数时候并不需要其他的模块,可能只需要一个Cyber,那么其实可以比较简单的编译cyber的。这篇教程就是教大……

阅读全文

cron创建定时任务

本文介绍 cron创建定时任务 cron创建定时任务 This article was original written by Jin Tian, welcome re-post, first come with https://jinfagang.github.io . but please keep this copyright info, thanks, any question could be asked via wechat: jintianiloveu cron是linux自带的一个定时任务管理器,通过cron创建定时任务需要如下步骤: 首先得把cron的log打开,不然你看不到log sudo vim /etc/rsyslog.d/50-default.conf 把cron.*那一行注释去掉。 然后……

阅读全文

LightHeadRCNN实现细节思考与Debug

本文介绍 LightHeadRCNN实现细节思考与Debug LightHeadRCNN实现细节思考与Debug This article was original written by Jin Tian, welcome re-post, first come with https://jinfagang.github.io . but please keep this copyright info, thanks, any question could be asked via wechat: jintianiloveu lightheadRCNN当中存在一个Proposal模块,这个与SSD是不同的。最近在训练LightHeadR……

阅读全文