probie的博客

训练流程：设置各种超参数定义数据加载模块 dataloader定义网络 model定义损失函数 loss定义优化器 optimizer遍历训练数据，预测-计算loss-反向传播对模型预测进行后处理测试流程：读取一张图片预处理模型预测对模型预测进行后处理后处理：目标框信息解码NMS非极大值抑制#######此代码只包含单个图像后处理，批处理参考原文####### Decode object coordinates from the form we regres

Stabilizing neural style-transfer for videoNeural Style Transfer on Real Time Video (With Full implementable code)Real Time Video Neural Style TransferStyle Transfer Guide

import cv2import numpy as npimg = cv2.imread('test_img/6_1.jpg')cv2.imshow("original", img)img_hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)rows, cols, channels = img.shape# 区间1lower_red = np.array([0, 43, 46])upper_red = np.array([10, 255, 255])mas

Solution:import numpy as npfrom PIL import Imagewith open('/media/yuxijin/dataset/videos/content/mountain_2/flow/reliable_9_10.txt', 'r') as f: data = f.readlines()header = list(map(int, data[0].split(' ')))w = header[0]h = header[1]vals = np.

If you are familiar with python, you can convert your ‘*.flo’ files to png and even video by tools provoided by georgegach/flowiz. Just follow their instruction, you can easily visualize your flow files.Below shows an example of the result by the python c

import globimport cv2import osfiles = glob.glob('bamboo_2/floviz/png/*.png')print(files)files.sort(key=lambda x: x[15:-4])files.sort()print(files)frame_list = []for i in files: img = cv2.imread(i) frame_list.append(img)size = (frame_li

MATLAB hsv颜色空间理解

psnr ssim 图像质量评估

图像锐化中高提升滤波的实现。

图像锐化中Laplacian锐化的实现

图像锐化中的Sobel锐化实现

自适应中值滤波的实现

图像平滑中的高斯平滑的实现

图像平滑中的平均平滑的实现

vs实现《数字图像处理与机器视觉》一书中图像旋转变换

vs实现《数字图像处理与机器视觉》一书中图像缩放变换

vs实现《数字图像处理与机器视觉》一书中图像平移变换

vs实现《数字图像处理与机器视觉》中图像的镜像变换，包括水平镜像和垂直镜像。

vs实现《数字图像处理与机器视觉》图像转置变换