Online to find a lot of people have emerged this mistake.
There are two reasons, one is not in accordance with the libGL library, then the installation: apt-get install libgl1-mesa-dev sudo is installed, but the path is not right, then the configuration of the path:"
LocatelibGL/usr/lib/i386 - Linux - gnu/mesa/libGL.so.1/usr/lib/i386 - Linux - gnu/mesa/libGL.so.1.2.0...
This is course of a job search algorithm. At the time of the study about Tkinter, then wrote a very simple robot maze of interface, and the use of the various search algorithms to search, as shown below:
Using A* to find the optimal path:
Because of the time, do not analyze, I put it on my own. Hope that some of the people who want to use Tkinter to help. Tkinter import from *
Random import from *
Numpy as import...
In the last part of the 1 part, we have analyzed the scaling and rotation of the feature points, and finally analyzed the method to remove the bad feature points. 2 the final step analysis of the vote of the basic idea is that these feature points relative to the center of the relative distance after the rotation and scaling into consideration is relatively constant, is to sense a frame feature points location relative to the center of the is unchanged. However, due to changes in the image itself, can not get exactly the same relative position, this time, there are some will be close to the center, there will be a lot of deviation. Then, the author uses the clustering method, chooses the biggest one kind as the best...
1 Introduction in the last blog, we analyzed the overall algorithm of CMT process and the implementation of the first few steps, then we continue to analyze the following steps. 2 Step 4, 5, 6 feature points matching and data fusion is a few steps through the tracking and feature matching to get the feature points of this frame, the two together. The last article analyzed the optical flow, here to analyze the characteristics of the match. The source code is as follows: keypoints compute, descriptors //Detect calculation of the key points of the current image
1 Introduction in the last article, I made a preliminary introduction to the CMT algorithm, and a preliminary analysis of the CppMT code, in this article, I will combine the author's thesis more comprehensive and detailed analysis of CMT algorithm. Here to explain the first, the author on the CMT algorithm to write two articles:
Matching and Consensus-based Tracking of Keypoints for Object Tracking (best paper reward wacv2014)...
1 introduction the full name of this algorithm is of Static-Adaptive Correspondences for Deformable Object Tracking Clustering, the article published in the CVPR2015, the official website for the:
Http://www.gnebehay.com/cmt/ this author is the author of the C++ version of the OpenTLD, including the OpenTLD version of the ROS is also based on his code to change the.
For CMT this trace count...
1 introduction last year I had to do a period of time the iPhone and LEGO ev3 robot research, had already developed IOS ev3 WiFi version of the SDK, can make the IOS and ev3 interact, but this is far short of the effect I've ever imagined, and look at my previous blog:
[EV3 and iOS hybrid robot programming one of the series] iOS what to do? EV3 can be more cool!
1 development environment 12.04ROS Hydro2 Ubuntu premise can refer to this article blog:http://prog3.com/sbdm/blog/yake827/article/details/44564057
Blog:http://prog3.com/sbdm/blog/celesius/article/details/39188119 has installed package adrone_autonomy and can run https://githu...
Visual tracking is one of the important sub problems in the field of computer vision CV. From my point of view, visual tracking is used on the robot, used in mobile devices, so why not put some tracking algorithm on the iPhone to see the actual tracking results. This is the most real contrast, the use of some video, after all, is not practical, and the key is not very good contrast in real time. Real time is the most important for mobile devices. The reason why there is a tracking problem, but also because the vast majority of the object detection Detection Object algorithm is too slow, simply can not be used in the...
1 Introduction in the last chapter of Blog, I introduced some methods of running CNN on iOS. But, in general, we need a strong performance machine to run CNN, we just need to get the results for the mobile terminal. Before the use of UFLDL in the Matlab code to modify the code to run a hand type identification of the 3 tier CNN, here we will consider the Matlab to C after transplantation into the xcode. 1:Matlab C Step first to ensure that the code can run, you can run, for example, my side, the following test CNN recognition hand:
1 introduction as a former iOS developer, in the study of deep learning, there is always an idea is to run on the iPhone deep learning, whether it is on the phone or the use of training good training data to test.
Because iOS development environment to support C++, so as long as your code is C/C++, in essence, you can run on the iOS.
How can we run CNN faster and better on iOS? 2 method 1: through the Matlab transcoding Matlab comes to turn into C tool, if you study the deep learning UFLDL tutorial, you know how to...
Self: is http://cs231n.github.io/python-numpy-tutorial/NumpyNumpy the core library for scientific computing in Python. It provides a high-performance multidimensional array object, and tools for working w...
Foreword this tutorial integrated CS231N UC and Berkerley CS188 Python Stanford tutorial.
The tutorial is very short, but for a certain basic programming language, learn other children's shoes. Python start python Python interpreter can have two kinds of use, a kind of is to use the interpreter interpreter, similar to Matlab, enter a line of code, run one; another is to write a py suffix documents, scripts, and python xxx.py running script scrip...
Here to share how to build on MAC, latex editing environment first, of course, is Google MAC latex for the existing blog tutorial, and found MAC common sublime text to edit, so the search MAC sublime latex, is found in the following two page:
Learning for Robotics Deep all kinds of resources summary
Contains Learning Computer, Vision Deep, Robotics, Learning Reinforcement...
1 foreword Python has many IDE to be able to use, the official oneself also brought one, Eclipse also can. But I use a variety of IDE, found that with the Text Sublime is the best use of A. Therefore, I have used Text Sublime to write Python program.
Then the whole building is actually very simple. Taking into account the next to do is scientific research, we also installed numpy, SciPy, Matplotlib and other plug-ins. 2 development environment Pro Macbook
OS X Mac 10...
1 Introduction currently there are more and more wearable devices using Bluetooth 4 BLE (Low Energy Bluetooth). For iOS development, Apple before the launch of the Framework CoreBluetooth to support the development of BLE. Hardware development for the understanding of the friend should know, before using the low version of Bluetooth devices, to connect to the iOS device, the need to register MFI, with the MFI protocol to carry out the corresponding development. If you are concerned about my previous research on EV3 LEGO, you can find...
At first, if you look at my first article about SDK's Blog DJI, I'm very excited about it. Unfortunately, after I deeply study the SDK and found the following unforgivable limit: 1) not enough GPS with not ground station and flight control function; 2) DJI directly interface can be used to control the size of phantom2 throttle. Phantom3 and inspire 1 seems to be able to control the throttle, but the take-off and landing is still restricted. More hateful is that SDK Phantom3 actually only supports Pr Phantom3...
1 introduction before this, the world has many research team spent a lot of time made four axis, to develop their own hardware, I finally can flight control research of Aerial Robotics, vision problems, great in a few years ago with the AR drone to do. But we know that the Drone AR flight range is too small, the camera is also very unstable, so use Drone AR to do research has a lot of limitations. Now, using DJI's aircraft, we have a very easy to have a very good research platform, to study the most cutting-edge issues, or that the most worthy of research questions, that is...
This article would like to introduce the basic construction of the SDK iOS DJI development environment, but DJI official has given a very detailed operation of its demo tutorials, web site as follows:
Here I summarize the steps to build a new project: Step 1: register a app
Here the identification code and app key is very important. DJI has done a limit here.
For iOS development, the identification code is the project identifier bundle. Key is used in the SDK...