The second parameter of Mat's CopyTo method is a mask. The length and width of the mask must be consistent with the source matrix. If the mask value of a pixel is not zero, it will be copied to the target matrix. When applied to an image, only part of the pixels can be copied by the mask to generate ...
Posted on Mon, 12 Aug 2019 05:25:34 -0700 by SuperCam
Recently, I have been researching PyTorch, which is a framework for in-depth learning. I want to use pytorch to realize the object detection of YOLOv3. During this process, I also read the articles written by my predecessors in various forums, posts, CSDN and so on. Thank you for helpi ...
Posted on Wed, 31 Jul 2019 18:18:56 -0700 by steveryan
In the process of traffic sign recognition, the adhering circular traffic signs are intercepted separately.
This is the effect after use, if the parameters are not used properly, the following situation will occur:
Usually, the drawing of the center and outline of the circle can be annotated whe ...
Posted on Sun, 28 Jul 2019 23:17:39 -0700 by kevingarnett2000
Read the documentation: Deep learning with PyTorch: 60 minutes quick start.
The goal of this tutorial is to:
Overall understanding of PyTorch's tensor library and neural network
Train a small neural network for image classification
What is PyTorch?
This is a Python-based scientific computing package that aims to:
Replace NumPy to play ...
Posted on Sat, 27 Jul 2019 14:49:06 -0700 by dylan001
After several days, it was finally installed. It's all pits that I've trampled on.
Maybe I am not proficient in using linux, and I don't know how to solve the problems in the configuration process. So I want to keep a comprehensive record.
I. Knowledge Points - Adding Environmental Variables
The M ...
Posted on Sun, 21 Jul 2019 01:49:31 -0700 by suzzane2020
When I first came into contact with opencv3, I found that there are too many uses of functions and C++ is not very good. I intend to record what I have learned so that I can check it later.
1. Read images: Mat imread (const string & filename, int flags);
The return value is Mat object
Mat type requires recording i ...
Posted on Thu, 04 Jul 2019 11:36:17 -0700 by dhorn
The general idea of target tracking is the key point in target tracking. TLD is also a tracking point (but not a key point such as tracking SIFT). Point tracking is based on optical flow method, specifically Pyramidal Lucas-Kanade tracker. This is a chance to introduce later. I recommend reading the ucas-Kanade Method section of Chapter 10 of ...
Posted on Sat, 29 Jun 2019 18:42:12 -0700 by franko75
The biggest difference between ImageNet data set and cifar, mnist data set is that the amount of data is very large; the size of a single picture is large, and the number of training samples is large; in the face of such a large data set, when converting into lmdb files, many new types of objects are used.
1. Dynamic expansion of the array &qu ...
Posted on Mon, 17 Jun 2019 13:19:20 -0700 by boinnk
1. Draw polygons directly with known boundary coordinates
2. Mask Generation by Morphological Operation
3. Human-computer interaction
In the field of image segmentation, an important task is to segment regions of interest (ROI). If it is a simple rectangular ROI region, it is very easy to segment. opencv's official pyt ...
Posted on Wed, 29 May 2019 04:59:07 -0700 by EODC
In the previous section This paper introduces how to apply YOLO to image target detection. After learning how to detect a single image, we can also use YOLO algorithm to achieve target detection in video stream.
Application of YOLO in Object Detection of Video Stream
First, open the yolo_video.py file and insert the following code:
# import the ...
Posted on Fri, 17 May 2019 13:58:50 -0700 by tim