[screenshot included] Error out of the box, while trying to predict 1.jpg in images directory, while using saved weights.
JordanMicahBennett opened this issue · 2 comments
See error above.
In fact, looking in the config file, there is no "backend' parameter in model object:
How to resolve?
Okay, after a few minutes across the various code files, I resolved the backend key error above, by placing backend = "Full Yolo" in the config model object.
Now a new error pops up, as it is looking for "full_yolo_backend.h5" as seen in screenshot below.
Edit: I found the missing yolo weights elsewhere on yolo page. I shall download, put in dir, and report later.
Okay, all issues fixed.
-
Modified config.json, to include "backend":"Full Yolo" in model object. (This was missing from config.json in main repository, and lead to a run-time error.)
-
To remove full_yolo_backend error, I downloaded full_yolo_backend.h5 from another repository to directory of this downloaded repository.
-
Successfully ran code on this random pic with potholes.
Test data from this random pic with potholes:
Input:
Result:
