image resizing

In early of this week, I found that each image was not resized. So, I thought the area value and convolution values could be wrong as each image resolution is different.

 

So, I applied resizimg logic for which I should change much even design.

 

By the way, I found the result of the work was worse than before. And my previous logic already reflect relative adjustment with different resolution.

 

I realized resizing image resolution could be harmful and make object recognition difficult.

 

Even if I spent much time in tight and busy schedule, I have learned the important lesson.

 

Yesterday, I recovered it to the previous one and I could fixed some bugs and improve performance better.

 

Of course, output data is not exact still. I should check it continuously.