kelicious/faster_rcnn

An implementation of Faster R-CNN applied to vehicle detection.

faster_rcnn's Issues

ImportError: cannot import name 'WEIGHTS_PATH_NO_TOP' from 'keras.applications.resnet50' (C:\ProgramData\Anaconda3\lib\site-packages\keras\applications\resnet50.py)

An error comes out when i use "annotate_video.py"

We are learning the network configuration about faster-rcnn currently, so i find out this great job you did. Right now, i have trained the data only from PASCAL VOC2007 based on resnet101 on Google Cloud Platform. After this, i got the result, but only on terminal and tmpout. So, i also want to give some images using annotate_video.py in order to plot the location box on these images from VOC2007, but i suffered from this error:

ValueError: operands could not be broadcast together with shapes (15390,) (30780,)

So, could you please help me out? Or could you please teach me how to plot the box to show the location of objects detected? I am quite new on deep learning. Thanks for your help and great work you did~

Performance?

Could you add some details to the README about what kind of AP performance your implementation was able to achieve for each dataset and object class?