I am outputting the bounding box coordinates by adding : printf("x_center: %d , y_center: %d , width: %d , height: %d \n",(right + left) / 2, (bot - top) / 2, right - left, bot - top); in image.c
However when I plot the center in my file, the y coordinate is biased (the x coordinate is ok). I'm not sure how to fix this, or the reasons for this result. Here is an example :
x_center: 1375 , y_center: 57 , width: 147 , height: 114. However I visually see that the target has a y_center of 540. My image is of size 3616 × 3040. The saved image prediction.jpg puts the bounding box at the right location !
There's probably a bias that is not corrected somewhere due to the network input size. Any idea ?
Hello,
I am outputting the bounding box coordinates by adding :
printf("x_center: %d , y_center: %d , width: %d , height: %d \n",(right + left) / 2, (bot - top) / 2, right - left, bot - top);in image.cHowever when I plot the center in my file, the y coordinate is biased (the x coordinate is ok). I'm not sure how to fix this, or the reasons for this result. Here is an example :
x_center: 1375 , y_center: 57 , width: 147 , height: 114. However I visually see that the target has ay_centerof 540. My image is of size 3616 × 3040. The saved image prediction.jpg puts the bounding box at the right location !There's probably a bias that is not corrected somewhere due to the network input size. Any idea ?
Thanks