![](https://img-blog.csdnimg.cn/20210511193753948.jpg?x-oss-process=image/resize,m_fixed,h_224,w_224)
计算机视觉
计算机视觉
physicist4AI
自然语言处理三大顶会审稿人 zhihu.com/people/guotong1988
展开
-
opencv python 入门实例
import cv2import numpy as npimg=np.ones((512,512,3), np.uint8)*255cv2.circle(img,center=(256,256), radius=128, color=(0,0,255), thickness=-1)cv2.circle(img,center=(256,256), radius=80, color=(...原创 2019-04-29 11:07:46 · 1450 阅读 · 0 评论 -
cv2 安装 | ImportError: No module named cv2
yum install opencv-python opencv-devel 然后下载opencv的zip,make了, 然后参考下面的方法做了 https://github.com/rbgirshick/fast-rcnn/issues/92 应该python就能import cv2了原创 2017-03-22 13:46:19 · 2381 阅读 · 0 评论 -
import cv2 报错 Segmentation fault (core dumped)
重新把python装了一遍 以及按照这个把opencv装了一遍 http://blog.csdn.net/daunxx/article/details/50506625原创 2017-01-11 20:15:32 · 3453 阅读 · 0 评论 -
OpenCV linux 安装
下载到tar包后yum install cmakecmake .makemake install原创 2015-11-25 20:36:58 · 757 阅读 · 1 评论 -
Masked Autoencoders 关键信息解读
摘要的最后一句话:achieves the best accuracy (87.8%) among methods that use only ImageNet-1K data. Transfer performance in downstream tasks outperforms supervised pre-training and shows promising scaling behavior.可见masked autoencoders (MAE) 的应用方法,是预训练好的MAE的encode原创 2022-05-13 16:10:03 · 269 阅读 · 0 评论 -
BA-NET 摘要速览
https://openreview.net/group?id=ICLR.cc/2019/ConferenceBA-NET: DENSE BUNDLE ADJUSTMENT NETWORKS本文介绍一个网络结构,来解决 structure-from-motion(SfM)问题,通过 特征矩阵 bundle adjustment (BA),这个网络结构能够明显的通过 特征矩阵误差 的形式 强化...原创 2020-03-16 10:14:02 · 285 阅读 · 0 评论 -
pix2pix 学习笔记
利用condition-GAN,以一图输入作为condition,生成一图,以完成image-to-image translation这个任务,而普通的GAN貌似只是噪声作为输入,原创 2019-04-12 10:24:42 · 867 阅读 · 0 评论 -
faster rcnn 基本概念
rpn:region proposal network IoU:Intersection-over-Union,交集并集之比 GT:ground truth,GT boxes (x1, y1, x2, y2, label),左上的坐标 和 右下的坐标 + 类别 im:image ROI:region of interest bbox:bounding-box regression:和on原创 2017-01-25 15:13:27 · 1099 阅读 · 0 评论 -
可视化comma.ai的h5图片数据
import cv2import numpy as npimport h5pyf = h5py.File('2016-06-08--11-46-01.h5', 'r')temp = f['X'][0] # 第0张img = temp.swapaxes(0,2).swapaxes(0,1)data = np.array(img[:,:,:])file = 'test.jpg'cv2.i原创 2017-01-13 14:54:56 · 1423 阅读 · 0 评论 -
可视化roadhackers的h5图片数据
import cv2import numpy as npimport h5pyf = h5py.File('135.h5', 'r')dset = f['1458552843.750']data = np.array(dset[:,:,:])file = 'test.jpg'cv2.imwrite(file, data)原创 2017-01-13 14:19:42 · 1587 阅读 · 0 评论