def process_sprites(sprite_paths_list, sprite_width, output_dir):
logger.info("Processing sprites")
processed_sprite_paths_list = []
for sprite_paths in sprite_paths_list:
processed_paths = []
for sprite_path in sprite_paths:
sprite_name = os.path.basename(sprite_path)
output_path = os.path.join(output_dir, sprite_name)
process_image(sprite_path, sprite_width, output_path)
processed_paths.append(output_path)
logger.debug("Processed sprite '" + sprite_name + "'")
processed_sprite_paths_list.append(processed_paths)
return processed_sprite_paths_list
def photo_ml(array):
image_age = image_sentiment = 0
image_emotions = {
"anger": 0,
"contempt": 0,
"disgust": 0,
"fear": 0,
"happiness": 0,
"neutral": 0,
"sadness": 0,
"surprise": 0
}
response = {}
for i in array:
url = i[
"image_url"] #'https://instagram.fsst1-2.fna.fbcdn.net/t51.2885-15/e35/17818239_241984309542803_360515474607308800_n.jpg'
text = i["content_text"]
if image.process_image(url) != False:
emotions = photo_processing.total_score(
image.process_image(url)[0])
info = image.process_image(url)[1]
age = sum([j["age"] for j in info["faces"]]) / len(info["faces"])
#sentiment_score = sentiment.process_text(text)
image_age += age
#if sentiment_score:
# image_sentiment += sentiment_score
image_emotions["anger"] += emotions[0]["anger"]
image_emotions["contempt"] += emotions[0]["contempt"]
image_emotions["disgust"] += emotions[0]["disgust"]
image_emotions["fear"] += emotions[0]["fear"]
image_emotions["happiness"] += emotions[0]["happiness"]
image_emotions["neutral"] += 0 #emotions[0]["neutral"]
image_emotions["sadness"] += emotions[0]["sadness"]
image_emotions["surprise"] += emotions[0]["surprise"]
n = len(array)
response['emotions'] = {
"anger": image_emotions["anger"],
"contempt": image_emotions["contempt"],
"disgust": image_emotions["disgust"],
"fear": image_emotions["fear"],
"happiness": image_emotions["happiness"],
"neutral": image_emotions["neutral"],
"sadness": image_emotions["sadness"],
"surprise": image_emotions["surprise"]
}
response['age'] = image_age / n
#response['sentiment']= image_sentiment/n
return response
def main():
#path = "flowers\\valid\\9\\image_06414.jpg"
#print(path)
#Image.open(path)
in_arg = get_input_args()
arch = in_arg.arch
checkpoint_path = in_arg.checkpoint
top_k = in_arg.top_k
#cat_to_name = in_arg.category_names
image_path = in_arg.image_path
#processing_unit = in_arg.gpu
if torch.cuda.is_available() and in_arg.gpu == 'gpu':
print('GPU will be used')
processing_unit = 'gpu'
elif torch.cuda.is_available() == False:
print('CPU will be used')
processing_unit = 'cpu'
with open(in_arg.category_names, 'r') as f:
cat_to_name = json.load(f)
#print(in_arg,arch,checkpoint_path,top_k,cat_to_name,processing_unit,image_path)
class_to_idx = torch.load(checkpoint_path)['class_to_idx']
model_new = pretrained_model(arch)
model = load_checkpoint(checkpoint_path, model_new)
image = process_image(image_path)
pros, indexes = predict(image, model, top_k, processing_unit)
check_predict(class_to_idx, cat_to_name, pros, indexes, top_k)
def predict(image_path, model, topk, gpu):
''' Predict the class (or classes) of an image using a trained deep learning model.
'''
image = torch.Tensor(process_image(image=image_path)).unsqueeze_(0)
device = torch.device(
"cuda" if gpu and torch.cuda.is_available() else "cpu")
model.to(device)
image = image.to(device)
model.eval()
with torch.no_grad():
logps = model.forward(image)
ps = np.exp(logps)
probs, classes = ps.topk(topk, dim=1)
return np.asarray(probs[0]) * 100, np.asarray(classes[0])
def predict(image_path, model, use_gpu, idx_to_class, topk):
if use_gpu:
model.to('cuda')
model.eval()
softmax = nn.Softmax(dim=1)
inputs = np.array([image.process_image(image_path)])
inputs = torch.from_numpy(inputs).float()
if use_gpu:
inputs.to('cuda')
outputs = model(inputs)
outputs = softmax(outputs)
topk = outputs.topk(topk)
probs = topk[0].to('cpu').detach().numpy()[0]
indices = topk[1].to('cpu').detach().numpy()[0]
del inputs, outputs
torch.cuda.empty_cache()
classes = [idx_to_class[str(x)] for x in indices]
return probs, classes
dest='y',
help='The new y images size',
required=True)
parser.add_argument(
'-s',
'--save_box_images',
dest='save_box_images',
help=
'If True, it will save the resized image and a drawed image with the boxes in the images',
default=0)
IMAGE_FORMATS = ('.jpeg', '.png', '.jpg')
args = parser.parse_args()
create_path(args.output_path)
create_path(''.join([args.output_path, '/boxes_images']))
args.dataset_path = add_end_slash(args.dataset_path)
args.output_path = add_end_slash(args.output_path)
for root, _, files in os.walk(args.dataset_path):
output_path = os.path.join(args.output_path, root[len(args.dataset_path):])
create_path(output_path)
for file in files:
if file.endswith(IMAGE_FORMATS):
file_path = os.path.join(root, file)
process_image(file_path, output_path, int(args.x), int(args.y),
args.save_box_images)
required=False)
parser.add_argument(
'-s',
'--save_box_images',
dest='save_box_images',
help=
'If True, it will save the resized image and a drawed image with the boxes in the images',
default=0)
IMAGE_FORMATS = ('.jpeg', '.JPEG', '.png', '.PNG', '.jpg', '.JPG')
args = parser.parse_args()
create_path(args.output_path)
if int(args.save_box_images):
create_path(''.join([args.output_path, '/boxes_images']))
args.dataset_path = add_end_slash(args.dataset_path)
args.output_path = add_end_slash(args.output_path)
for root, _, files in os.walk(args.dataset_path):
output_path = os.path.join(args.output_path, root[len(args.dataset_path):])
create_path(output_path)
for file in files:
if file.endswith(IMAGE_FORMATS):
file_path = os.path.join(root, file)
process_image(file_path, output_path, args.x, args.y,
args.save_box_images, args.mode)
def hello():
if request.method == 'GET':
bucket_name = request.args.get('bucket')
key = request.args.get('key')
res = process_image(bucket_name, key)
return jsonify(res)
# '-m',
# '--margins',
# dest='margins',
# help='...',
# default=None
# )
IMAGE_FORMATS = ('.jpeg', '.JPEG', '.png', '.PNG', '.jpg', '.JPG')
args = parser.parse_args()
args.dataset_path += '/data_train'
args.output_path += '/data_train'
create_path(args.output_path)
create_path(''.join([args.output_path, '/boxes_images']))
args.dataset_path = add_end_slash(args.dataset_path)
args.output_path = add_end_slash(args.output_path)
assert os.path.exists(
args.dataset_path
), 'All data (images and annotations) should be within a folder named \'data_train\''
for root, _, files in os.walk(args.dataset_path):
output_path = os.path.join(args.output_path, root[len(args.dataset_path):])
create_path(output_path)
for file in files:
if file.endswith(IMAGE_FORMATS):
file_path = os.path.join(root, file)
process_image(file_path, output_path, int(args.w), int(args.h),
args.save_box_images, args.do_crop, args.do_sub_crop)
