Deeplab v3 predicts only with default ratio 512 * 512. how to process images with different size ratio.
please look over the error:-
Traceback (most recent call last):
File "", line 35, in predicted = model_dlv3.predict(img[np.newaxis, ...])
File "/home/developer/anaconda2/envs/py36/lib/python3.6/site-packages/keras/engine/training.py", line 1149, in predict x, _, _ = self._standardize_user_data(x)
File "/home/developer/anaconda2/envs/py36/lib/python3.6/site-packages/keras/engine/training.py", line 751, in _standardize_user_data exception_prefix='input')
File "/home/developer/anaconda2/envs/py36/lib/python3.6/site-packages/keras/engine/training_utils.py", line 138, in standardize_input_data str(data_shape))
ValueError: Error when checking input: expected input_3 to have shape (512, 512, 3) but got array with shape (2448, 3264, 3)
You can resize the images to 512*512 or turn images into a square using padding and then resize them into 512*512. Latter option is better since it won't change the shape of the objects inside the image. You can use different padding modes and see what gives the best result.
You can use opencv or scikit-image for resizing and padding
opencv documentation for padding: https://docs.opencv.org/2.4/modules/imgproc/doc/filtering.html?highlight=copymakeborder#copymakeborder
scikit-image documentation for padding : http://scikit-image.org/docs/dev/api/skimage.util.html#skimage.util.pad