Added environment.yml. Updated README to reflect using environment.ym…

…l. Updated code to work with tf2. Removed some unused dependencies.

README.md

@@ -25,32 +25,29 @@ An overview of predictions of biotic and anthropogenic activity on recordings of
 
 ## Requirements
 
-The system has been tested using:
+The system has been tested using the dependencies in `environment.yml`. Our code works with python 3.
 
-- `Ubuntu 16.04`
-- `python 2.7`
-- The `anaconda` distribution (see [here](https://www.anaconda.com/download/) for download details).
-
-Run the following commands to get suitable versions of the required libraries:
-
-    pip install -r https://raw.githubusercontent.com/Lasagne/Lasagne/v0.1/requirements.txt
-    pip install Lasagne==0.1
-    pip install --upgrade https://github.com/Lasagne/Lasagne/archive/master.zip
-    pip install nolearn
-    pip install librosa
-    pip install easydict
-    pip install tqdm
-    pip install git+git://github.com/mdfirman/ml_helpers.git@master
-
-For training and testing we used a 2GB NVIDIA GPU. The computation requirements for classification are pretty low though, so a GPU should not be required.
+You can create an environment with all the dependencies installed using:
 
+```bash
+conda env create -f environment.yml -n citynet
+conda activate citynet
+```
 
 ## How to classify a new audio file with CityNet
 
 - Run `python demo.py` to classify an example audio file.
 - Predictions should be saved in the folder `demo`.
 - Your newly-created file `demo/prediction.pdf` should look identical to the provided file `demo/reference_prediction.pdf`:
 
+## How to classify multiple audio files
 
+You can run CityNet on a folder of audio files with:
 
-Editing `demo.py` should allow you to classify your own audio files.
+```bash
+python multi_predict.py path/to/audio/files
+```
+
+## Hardware requirements
+
+For training and testing we used a 2GB NVIDIA GPU. The computation requirements for classification are pretty low though, so a GPU should not be required.

demo.py

@@ -3,7 +3,7 @@
 import zipfile
 import numpy as np
 import pickle
-from six.moves import urllib
+import urllib
 import matplotlib.pyplot as plt
 import sys
 sys.path.append('lib')
@@ -42,7 +42,7 @@
 for classifier_type in ['biotic', 'anthrop']:
 
     with open('tf_models/%s/network_opts.yaml' % classifier_type) as f:
-        options = yaml.load(f)
+        options = yaml.full_load(f)
 
     model_path = 'tf_models/%s/weights_99.pkl-1' % classifier_type
 

lib/train_helpers.py

@@ -1,11 +1,8 @@
 import numpy as np
 import os
-import sys
-import yaml
 import collections
-import tensorflow as tf
-from tensorflow import layers
-from tensorflow.contrib import slim
+import tensorflow.compat.v1 as tf
+import tf_slim as slim
 import minibatch_generators as mbg
 
 
@@ -125,6 +122,8 @@ def __iter__(self): ##, num_per_class, seed=None
 def create_net(SPEC_HEIGHT, HWW_X, LEARN_LOG, NUM_FILTERS,
     WIGGLE_ROOM, CONV_FILTER_WIDTH, NUM_DENSE_UNITS, DO_BATCH_NORM):
 
+    tf.compat.v1.disable_eager_execution()
+
     channels = 4
     net = collections.OrderedDict()
 

multi_predict.py

@@ -1,11 +1,7 @@
 import os
 import sys
 import yaml
-import zipfile
-import numpy as np
 import pickle
-from six.moves import urllib
-import matplotlib.pyplot as plt
 import sys
 sys.path.append('lib')
 
@@ -29,7 +25,7 @@
 for classifier_type in ['biotic', 'anthrop']:
 
     with open('tf_models/%s/network_opts.yaml' % classifier_type) as f:
-        options = yaml.load(f)
+        options = yaml.full_load(f)
 
     model_path = 'tf_models/%s/weights_99.pkl-1' % classifier_type
 

prediction/tf_classifier.py

@@ -1,12 +1,9 @@
-import os
 import sys
-import yaml
 import numpy as np
 from time import time
-import pickle
 from collections import namedtuple
 
-import tensorflow as tf
+import tensorflow.compat.v1 as tf
 import librosa
 from librosa.feature import melspectrogram
 from scipy.io import wavfile