/
main.py
60 lines (45 loc) · 1.77 KB
/
main.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
from flask import Flask, redirect, render_template, request
from gevent.wsgi import WSGIServer
import glob
import logging
import os
import sys
from predict import InstrumentClassifier
app = Flask(__name__)
app.config['MAX_CONTENT_LENGTH'] = 1 * 2**20
app.logger.addHandler(logging.StreamHandler(sys.stdout))
app.logger.setLevel(logging.ERROR)
model_id = '2016-10-15_22-11-47_31fdbcbb'
model_dir = 'static/model/' + model_id
model = InstrumentClassifier(model_dir)
print('Using model:', model_id)
example_files = [os.path.basename(file) for file in glob.glob('static/audio/*.flac')]
@app.route('/')
def hello():
return render_template('home.html', model_id=model_id, example_files=example_files)
@app.route('/classify/instrument', methods=['POST'])
def classify():
if 'audio_file' not in request.files:
return redirect('/')
# File-like object than can be directy passed to soundfile.read()
# without saving to disk.
audio_file = request.files['audio_file']
if audio_file.filename == '':
return redirect('/')
class_probabilities = model.predict_probabilities(audio_file)
class_probabilities = class_probabilities.round(5)
label = model.class_label_from_probabilities(
class_probabilities)
return render_template('home.html',
model_id=model_id,
example_files=example_files,
audio_file=audio_file.filename,
predicted_label=label,
class_probabilities=class_probabilities)
if __name__ == '__main__':
# - for local debugging run: python main.py && open http://localhost:5000/
# - for Heroku production run via the Procfile
app.debug = True
# needed since Flask dev mode interacts badly with TensorFlow
http_server = WSGIServer(('', 5000), app)
http_server.serve_forever()