cleanups for 1.1.0 release (#26)

Dockerfile

@@ -1,4 +1,4 @@
-FROM codait/max-base:v1.1.0
+FROM codait/max-base:v1.1.1
 
 ARG model_bucket=http://max-assets.s3-api.us-geo.objectstorage.softlayer.net/image-caption-generator/1.0
 ARG model_file=assets.tar.gz

README.md

@@ -1,6 +1,6 @@
 [![Build Status](https://travis-ci.org/IBM/MAX-Image-Caption-Generator.svg?branch=master)](https://travis-ci.org/IBM/MAX-Image-Caption-Generator)
 
-# IBM Developer Model Asset Exchange: Show and Tell Image Caption Generator
+# IBM Developer Model Asset Exchange: Image Caption Generator
 
 This repository contains code to instantiate and deploy an image caption generation model. This model generates captions from a fixed vocabulary that describe the contents of images in the [COCO Dataset](http://cocodataset.org/#home). The model consists of an _encoder_ model - a deep convolutional net using the Inception-v3 architecture trained on [ImageNet-2012 data](http://www.image-net.org/challenges/LSVRC/2012/) - and a _decoder_ model - an LSTM network that is trained conditioned on the encoding from the image _encoder_ model. The input to the model is an image, and the output is a sentence describing the image content.
 

config.py

@@ -18,12 +18,13 @@
 VOCAB_FILE = './assets/word_counts.txt'
 # for image models, may not be required
 MODEL_INPUT_IMG_SIZE = (299, 299)
-MODEL_LICENSE = 'APACHE V2'
+MODEL_LICENSE = 'Apache 2.0'
 
 MODEL_META_DATA = {
-    'id': '{}-tensorflow'.format(MODEL_NAME.lower()),
-    'name': '{} TensorFlow Model'.format(MODEL_NAME),
+    'id': API_TITLE.lower().replace(' ', '-'),
+    'name': API_TITLE,
     'description': '{} TensorFlow model trained on MSCOCO'.format(MODEL_NAME),
-    'type': 'image_captioning',
-    'license': '{}'.format(MODEL_LICENSE)
+    'type': 'Image-to-Text Translation',
+    'license': MODEL_LICENSE,
+    'source': 'https://developer.ibm.com/exchanges/models/all/max-image-caption-generator/'
 }

tests/test.py

@@ -23,26 +23,34 @@ def test_metadata():
     assert r.status_code == 200
 
     metadata = r.json()
-    assert metadata['id'] == 'im2txt-tensorflow'
-    assert metadata['name'] == 'im2txt TensorFlow Model'
+    assert metadata['id'] == 'max-image-caption-generator'
+    assert metadata['name'] == 'MAX Image Caption Generator'
     assert metadata['description'] == 'im2txt TensorFlow model trained on MSCOCO'
-    assert metadata['license'] == 'APACHE V2'
+    assert metadata['license'] == 'Apache 2.0'
+    assert metadata['type'] == 'Image-to-Text Translation'
+    assert 'max-image-caption-generator' in metadata['source']
 
 
-def test_predict():
-    model_endpoint = 'http://localhost:5000/model/predict'
-    file_path = 'assets/surfing.jpg'
+def _check_response(r):
     caption_text = 'a man riding a wave on top of a surfboard .'
-
-    with open(file_path, 'rb') as file:
-        file_form = {'image': (file_path, file, 'image/jpeg')}
-        r = requests.post(url=model_endpoint, files=file_form)
-
     assert r.status_code == 200
     response = r.json()
     assert response['status'] == 'ok'
     assert response['predictions'][0]['caption'] == caption_text
 
 
+def test_predict():
+    model_endpoint = 'http://localhost:5000/model/predict'
+    formats = ['jpg', 'png']
+    file_path = 'tests/surfing.{}'
+
+    for f in formats:
+        p = file_path.format(f)
+        with open(p, 'rb') as file:
+            file_form = {'image': (p, file, 'image/{}'.format(f))}
+            r = requests.post(url=model_endpoint, files=file_form)
+        _check_response(r)
+
+
 if __name__ == '__main__':
     pytest.main([__file__])