CarND-Vehicle-Detection
CarND-Vehicle-Detection copied to clipboard
xslittlegrass
No bounding boxes
I downloaded the weights from http://pjreddie.com/media/files/yolo-tiny.weights but after I run the script, no bounding boxes are seen on the test images.
I tried the weight file from the link in the readme (the link directing to google drive 172MB ). But there is no box! When I assigned the threshold=0, I can see all boxes.
Had the same problem, no boxes produced. I used the weight file from the link in the readme. I noticed in the notebook
import keras # broken for keras >= 2.0, use 1.2.2 My keras is 2.0.6, but No running errors reported, on a Mac OS X
After changing keras to 1.2.2, still no boxes produced, no error messages.
Any suggestions to find out why? Thanks
When I assigned the threshold=0, I can see all boxes.
The same happened to me too, when threshold=0, there are many boxes around the areas of the vehicles.
It would be great to be able to visualize the out array to show the 7x7 grid, BBox candidates, and probabilities.
same problem here i can see bounding boxes only when the threshold is reduced to less than 0.005 but it is outside the car....
I am also facing same issue, when I change the threshold to 0 it draws many boxes
same issue both with the test images as well as test video. Any update how to solve this ?
@trohit920 @iamsome1 @tianqig @declanzane @eserercanakin follow this this worked for me i think this post is old but ... yes it works
following worked for me
comment by mark-chan https://github.com/xslittlegrass/CarND-Vehicle-Detection/issues/3
add this code before the load_weights(model,'./yolo-tiny.weights') function...
@anandkoirala thanks a lot it solves the problem of bounding boxes, but why i should put it before the load_weights() function. when i put in the utils module it doesn't work.
@AhmedAAkl so that this function you have pasted will be called not the one from the utils file. (they have the same name). you can run the function from util file if the original load_weights is replaced by new one.
@trohit920 @iamsome1 @tianqig @declanzane @eserercanakin follow this this worked for me i think this post is old but ... yes it works following worked for me comment by mark-chan #3 add this code before the load_weights(model,'./yolo-tiny.weights') function...
@anandkoirala @wizholy Hello, I know this post is old, nevertheless I would greatly appreciate your help :-) I've tried to use the alternative function for loading the weights, as suggested in the referenced, still, the problems remain the same: No bounding boxes, unless one lower the threshold almost to zero...
I read that it did work for someone here...Any help is very very welcome!
