The first experience of scientific research

Ashamed to say, the second grade of graduate students, for the first time the entire scientific process has a general understanding.

Thanks to the computer vision algorithm and the application of this book. The following appendix contains a lot of resources that can be used, and what I do is the 14th chapter of the boost algorithm for pedestrian detection.

Http://people.csail.mit.edu/torralba/shortCourseRLOC/boosting/boosting.html

It's almost no problem to follow the steps of the experiment.

There are several main points:

1. There are no steps to establish a database in the tutorial, you need to download LabelMe toolbox. INIT.M modified as follows

Addpath (' E:\Download\boostingDemo ')
Addpath (' E:\Download\LabelMeToolbox\LabelMeToolbox ')%toolbox the decompression path to find the function
Addpath (' Tools ')
Addpath (' Gentleboost ')

The path of the PARAMATERS.M should also be modified accordingly

2. There are not many functions found in the first experiment, it is possible to copy the functions in Genboost to the current directory, but it should be unnecessary, possibly because the path is not well-set.

3. There is also in the experimental process in strict accordance with the operating procedures. There will be no error. Some need to wait for a long time, such as COMPUTEFEATURE.M. Do not tamper with the keyboard during operation, or make an error. And the error of the program will make you think it is the problem of the program itself. Not really.

4. Without an accident, you can fully experience the characteristics of the calculation, training, testing, precision recall curves and so on in many papers convoluted things. Screenshot see.

5. From the process of the experiment, it can be found that not every experiment is to implement each function from the beginning, completely can be the code in other people's papers, or other tool (some. m files), such as the boost algorithm, such as LabelMe in the program to invoke. This book lists many test datasets and toolset. This should be a necessary thing for scientific research. Take these things slowly and use them. Research is almost a malicious primer.

6. Take a look at this MATLAB code. Know its essence after knowing its appearance. Not much, but in fine. A pass and hundred pass. Learn MATLAB programming, make visual need, use of MATLAB.