def preprocess_visual_features_for_jsons(metajson_files, downloaded_images_dir=None):
for metajson_file in metajson_files:
metadata = parse_json_file(metajson_file)
if metadata == None:
print "Could not read json file %s" % metajson_file
continue
print_status("ID: " + metadata["id"] + " File name: " + metajson_file + "\n")
if metadata["stat"] == "ok":
try:
if not downloaded_images_dir == None:
image_filename = metajson_file.split(os.sep)[-1].replace('.json', '.jpg')
image_path = downloaded_images_dir + os.sep + image_filename
image = Image(image_path).toHSV()
else:
url = get_small_image_url(metadata)
image = Image(url).toHSV()
except Exception:
print "Could not get image:", metadata["id"]
continue
visual_data = {}
visual_data = extract_colors(image, visual_data, 5)
visual_data = extract_edges(image, visual_data, 5)
file_name_for_visual_metadata = metajson_file.replace('.json', '_visual.json')
write_json_file(visual_data, file_name_for_visual_metadata)
else:
print "Status was not ok:", metadata["id"]
def extract_features(image_cluster, metadata_dir):
images = []
for metajson_file, _ in image_cluster:
relative_path_to_json = construct_path_to_json(metajson_file)
full_path_to_json = metadata_dir + relative_path_to_json
metadata = parse_json_file(full_path_to_json)
if metadata == None:
continue
if metadata["stat"] == "ok":
data = {}
url = get_small_image_url(metadata)
data["image_id"] = metadata["id"]
data["file_path"] = metajson_file
data["url"] = url
try:
image = Image(url).toHSV()
except Exception:
continue
data = extract_colors(image, data, 5)
data = extract_edges(image, data, 5)
images.append(data)
return images
