roachsinai

void imageScaleAndRotate(const cv::Mat &src, cv::Mat &out, double scale, double roll) {
    // scaling
		cv::Mat imSmall;
		cv::resize(src, imSmall, cv::Size(), scale, scale);
    
    // prepare for rotating
		int width = imSmall.cols, height = imSmall.rows;
		int diagonal = int(sqrt(height * height + width * width));

		int offsetX = (diagonal - width) / 2, offsetY = (diagonal - height) / 2;

roachsinai

Table Of Content

Skip to the relevant sections if needed.

• 2-min tutorial to do it the quick-and-dirty-way
• Concepts for resolving Git conflicts
• Setting up different editors / tool for using git mergetool
  • Finding out what mergetool editors are supported
• mergetool simple code example for vimdiff
  • Resolving conflict from a Git pull
• Other great references and tutorials

roachsinai

#include <cstddef>
#include <cstdlib>

template <typename R, auto getter, auto setter, size_t (*offset)()>
struct property {
    inline R *self() {
        return reinterpret_cast<R *>(reinterpret_cast<size_t>(this) - offset());
    }
    inline operator auto() { return (self()->*getter)(); }
    inline void operator=(auto t) { (self()->*setter)(t); }

roachsinai

all:
	cc -shared -fPIC lib_one.c lib_one.h -o libone.so
	cc -shared -fPIC lib_two.c lib_two.h -o libtwo.so -L. -lone -Wl,-rpath=libs
	mkdir -p libs
	mv libone.so libs
	# -L. 是针对编译时的，使得链接器（ld）可以在 gcc 编译时找到 libtwo.so 让 name loopup 成功
	# -rpath=. 是针对运行时的，使得在执行 main 时让动态链接器（ld.so）可以找到 main 依赖的 so
	# Wl 表示后面的是一个链接器选项（针对 ld，而不是 gcc 的选项）
	cc main.c -o main -L. -ltwo -Wl,-rpath=.

roachsinai

/* recursive mkdir based on
http://nion.modprobe.de/blog/archives/357-Recursive-directory-creation.html
*/
#include <unistd.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <string.h>

#define PATH_MAX_STRING_SIZE 256

roachsinai

#!/bin/bash
## This gist contains instructions about cuda v11.2 and cudnn8.1 installation in Ubuntu 20.04 for Pytorch 1.8 & Tensorflow 2.7.0

### steps ####
# verify the system has a cuda-capable gpu
# download and install the nvidia cuda toolkit and cudnn
# setup environmental variables
# verify the installation
###

roachsinai

#include <assert.h>
#include <stdint.h>
#include <stdio.h>
#include <string.h>
#include <windows.h> // 各种位图数据结构

class Converter
{
public:
    Converter() : pixels_(NULL), width_(0), height_(0) {}

roachsinai

#!/bin/sh
# Copy the contents of this file to the clipboard, then get a terminal open on your device and enter:  
#   $ cat > n.sh  
#  [Ctrl+V] or Right Click, Paste. Then [Ctrl+D].
#  chmod +x n.sh
#  To run: ./n.sh eth0
SLP=1 # display / sleep interval
DEVICE=$1
IS_GOOD=0
for GOOD_DEVICE in `grep \: /proc/net/dev | awk -F: '{print $1}'`; do

roachsinai

name: "YOLONET"
layer {
  name: "data"
  type: "Input"
  top: "data"
  input_param { shape: { dim: 1 dim: 3 dim: 416 dim: 416 } }
}
layer {
  name: "conv1"
  type: "Convolution"

roachsinai

WORK IN PROGRESS

PyTorch Internals Part II - The Build System

In the first post I explained how we generate a torch.Tensor object that you can use in your Python interpreter. Next, I will explore the build system for PyTorch. The PyTorch codebase has a variety of components:

• The core Torch libraries: TH, THC, THNN, THCUNN
• Vendor libraries: CuDNN, NCCL
• Python Extension libraries
• Additional third-party libraries: NumPy, MKL, LAPACK