FabianGroeger96 / cnn-number-detection

Number detection implemented using TensorFlow with custom CNN architecture for fast inference and custom dataset
MIT License
60 stars 16 forks source link

imresize is not present in scipy.misc #4

Closed nlbutts closed 5 years ago

nlbutts commented 5 years ago

When I run the import lines in extract_data.py I get the following error message: ` In [5]: from Trainer.Models.model_gnet_deep_v2 import ModelGNetDeepV2 ...:
Using TensorFlow backend.

ImportError Traceback (most recent call last)

in ----> 1 from Trainer.Models.model_gnet_deep_v2 import ModelGNetDeepV2 ~/projects/cnn-number-detection/Trainer/Models/model_gnet_deep_v2.py in 1 import constants ----> 2 from Trainer.Models.model import Model 3 from keras.models import Sequential 4 from keras.layers import Dense, Dropout, Flatten 5 from keras.layers import Conv2D, MaxPooling2D ~/projects/cnn-number-detection/Trainer/Models/model.py in 18 from tensorflow.python.framework.graph_util import convert_variables_to_constants 19 from tensorflow.python.keras.preprocessing.image import ImageDataGenerator ---> 20 from vis.visualization import visualize_activation 21 from vis.visualization import visualize_saliency 22 from vis.visualization import visualize_cam, overlay ~/projects/cnn-number-detection/venv/lib/python3.6/site-packages/vis/visualization/__init__.py in 5 from .activation_maximization import visualize_activation 6 ----> 7 from .saliency import visualize_saliency_with_losses 8 from .saliency import visualize_saliency 9 from .saliency import visualize_cam_with_losses ~/projects/cnn-number-detection/venv/lib/python3.6/site-packages/vis/visualization/saliency.py in 3 import numpy as np 4 import matplotlib.cm as cm ----> 5 from scipy.misc import imresize 6 7 from keras.layers.convolutional import _Conv ImportError: cannot import name 'imresize' `
FabianGroeger96 commented 5 years ago

Try to install and let me know if it works: pip install scipy==1.1.0 There might be some missing packages in the requirements.txt

FabianGroeger96 commented 5 years ago

Closing this issue due to no response from issuer