When creating a post, please add:
Week3 Convolutional Neural Networks
lab1
have debugged the code again and again but still getting the same error.
Please post a screen capture image that shows the error or any assert messages.
There is an error in your code in the yolo_non_max_suppression() function. The “scores” return value is incorrect when the “classes” are set to all-zeros.
Be sure you are using the newest version of this notebook and the unit_tests.py file. Both were updated in the last three weeks.
I started the assignment today.
how can i update to the latest versions?
That is covered on this thread. Make sure to read all the way to the end to see the discussion about how to handle the supporting files in addition to the notebook (ipynb file).
If you just started the assignment, you should automatically have the newest version.
Be very careful that you don’t use any guidance for this assignment that you find online - because it won’t apply to the new version.
Tom is right, of course. There’s also an easy way to check: the current version is v2.0 as shown in this screenshot:
How to solve this problem? I came up with the same issue here. It seems that scores return two elements but the result suppose to be one. I am not familar with tensorflow.
But when I submit the coding test, I got 100 points. Quite Weird.
It would be worth seeing your code to understand what the bug is that causes the test to fail for the case of the same label. It may be a problem with the grader that it does not also catch that bug. We can’t share code in a public thread, but please check your DMs for a message from me about how to proceed with that.
I am having a similar error, have been debugging this for a while, not sure where I am making a mistake. Any advice would be appreciated.
Notice that the test case expected two output values for that test, but your code only generated one. The first thing to check is how you are calling the underlying TF function there. E.g. make sure you are not hard-coding any of the parameters.
Thank you! I walked through the code multiple times, trying to understand the flow between the classes, boxes label, and score label. I am not hard coding anything in the call to tf.image.non_max_suppression(). I’m not sure what else to check at this point.
Maybe it’s time to look at your code. We can’t do that on this public thread, but please check your DMs for a message from me about how to proceed. You can recognize DMs in your normal “feed” by the little envelope icon.
Thank you for your help! For anyone finding this post in the future, my issue was with the final output, I was not using the correct variable. Once that was fixed the code ran as expected.
3 Likes
Just wanted to add a comment for anyone else stuck on this. All of the code you need to implement needs to be inside the for loop. Not sure how my code got indented badly.
1 Like
There are several common errors on this function. One is hard-coding the iou_threshold in the TF NMS call. The other is getting the arguments in the wrong order on the call to tf.gather in the loop.