English/Chinese subtitles are available for this video.
有中文字幕描述！

Hi, all.
As I was working on bird detection. I need to prepare a dataset for training the SVM classifier. Here is how I did it. I use a small dataset (around 200 images), and trained the SVM with that one, then, I used the trained SVM as a gauge to measure how good a candidate new sample would be. As you can see from the video, the software tool will display the value of the distance the new image would be from the support vector. Any comment is appreciated.

The tool is open-sourced, you may find the code in GitHub: https://github.com/Chunde/Image-Cropp...