Closed jmd9019 closed 6 years ago
Try to input a dictionary instead of an integer into RCNN, e.g.
model = keras_rcnn.models.RCNN((224, 224, 3), ["circle", "rectangle", "triangle"])
Closing. @jmd9019 Let me know if @jhung0's comment did not resolve your issue.
I also apologize for the API instability. @jhung0 and I are in the process of (semi-) stabilizing the object detection API. I'll also look at stabilizing the image segmentation API soon.
@0x00b1 @jhung0 comment did not work still facing the same issue.
One more question is this issue is because of I am running on aws ec2 instance in jupyter??
No, that's not an issue. Do you have everything updated? Have you tried what's in the readme?
Yes everything is updated but haven't tried as given in readme. And I tried in Kaggle there was no issue. But I am getting this error when running on aws so i asked the above question
Is it really the same error? If you input a dictionary, I'm not sure why it would give you TypeError: object of type 'int' has no len()
No getting different error now
codes:
import keras from keras.layers import Input from keras_rcnn.models import RCNN
img_in = Input((None, None, 3)) rcnn_model = RCNN((225,225, 2),["circle", "rectangle", "triangle"]) optimizer = keras.optimizers.Adam(0.0001) rcnn_model.compile(optimizer) print('Input:', rcnn_model.get_input_shape_at(0)) print('Output:', rcnn_model.get_output_shape_at(0))
Error:
Running on jupyter with python 3.6 anaconda environment
TypeError Traceback (most recent call last)