def load_model(ROOT_DIR, device = 'GPU'):
# return 'abc', []
device_scope = '/{}:0'.format(device.lower())
with tf.device(device_scope):
MODEL_DIR = os.path.join(ROOT_DIR, 'demo', 'model', 'logs')
class InferenceConfig(mapillary.MapillaryConfig):
GPU_COUNT = 1
IMAGES_PER_GPU = 1
inference_config = InferenceConfig()
print (' - Batch Size : ', inference_config.BATCH_SIZE)
model = modellib.MaskRCNN(mode="inference", config=inference_config, model_dir=MODEL_DIR)
# model_path = model.find_last()[1]
model_path_base = '/home/play/playment/production/Mask_RCNN/demo/model/logs/mapillary20180315T0317/'
models = ['mask_rcnn_mapillary_0196.h5','mask_rcnn_mapillary_0158.h5', 'mask_rcnn_mapillary_0083.h5']
model_path = os.path.join(model_path_base, models[0])
print (' - Model Path : ', model_path)
if model_path != None:
model.load_weights(model_path, by_name=True)
return model, inference_config
else:
sys.exit(1)
return []
config = HarzConfig()
else:
class InferenceConfig(HarzConfig):
# Set batch size to 1 since we'll be running inference on
# one image at a time. Batch size = GPU_COUNT * IMAGES_PER_GPU
GPU_COUNT = 1
IMAGES_PER_GPU = 1
config = InferenceConfig()
config.display()
# Create model
if args.command == "train":
model = modellib.MaskRCNN(mode="training",
config=config,
model_dir=args.logs)
else:
model = modellib.MaskRCNN(mode="inference",
config=config,
model_dir=args.logs)
if args.weights and args.weights.lower() == "last":
# Find last trained weights
print("Loading model from previous weight file!")
weights_path = model.find_last()
model.load_weights(weights_path, by_name=True)
# Select weights file to load
# if args.weights.lower() == "coco":
# weights_path = COCO_WEIGHTS_PATH
class InferenceConfig(MyConfig):
GPU_COUNT = 1
IMAGES_PER_GPU = 1
DETECTION_MIN_CONFIDENCE = 0.5
config = InferenceConfig()
config.display()
for k, v in vars(args).items():
setattr(config, k, getattr(args, k))
config.display()
# Create model
if args.command == "train":
model = modellib.MaskRCNN(mode="training", config=config)
else:
model = modellib.MaskRCNN(mode="inference", config=config)
if config.weights:
# Find last trained weights
print("Loading model from previous weight file!")
if config.weights == "last":
weights_path = model.find_last()
model.load_weights(weights_path, by_name=True)
else:
model.load_weights(config.weights, by_name=True)
# Train or evaluate
if args.command == "train":
train(model, config)
if 'src.model' in sys.modules : del sys.modules['src.model']
if 'src.utils' in sys.modules : del sys.modules['src.utils']
if 'src.parallel_model' in sys.modules : del sys.modules['src.parallel_model']
import src.utils as utils
import src.model as modellib
import src.parallel_model as parallel_model
if 'keras.' in sys.modules : del sys.modules['keras']
import keras
MODEL_DIR = os.path.join(ROOT_DIR, 'demo', 'model', 'logs')
COCO_MODEL_PATH = os.path.join(ROOT_DIR, 'demo', 'model', "mask_rcnn_coco.h5")
model = modellib.MaskRCNN(mode="training", config=mapillary_config, model_dir=MODEL_DIR)
# model.keras_model = parallel_model.ParallelModel(model.keras_model, mapillary_config.GPU_COUNT, verbose = 0)
# model.keras_model.summary()
init_with = "coco" # imagenet, coco, or last
# init_with = "last" # imagenet, coco, or last
if init_with == "imagenet":
model.load_weights(model.get_imagenet_weights(), by_name=True)
elif init_with == "coco":
model.load_weights(COCO_MODEL_PATH, by_name=True,
exclude=["mrcnn_class_logits", "mrcnn_bbox_fc",
"mrcnn_bbox", "mrcnn_mask"])
elif init_with == "last":
model_path = model.find_last()[1]
print (' - Path : ', model_path)
all visualizations in the notebook. Provide a
central point to control graph sizes.
Change the default size attribute to control the size
of rendered images
"""
_, ax = plt.subplots(rows, cols, figsize=(size * cols, size * rows))
return ax
inference_config = InferenceConfig()
inference_config.display()
# Recreate the model in inference mode
model = modellib.MaskRCNN(mode="inference",
config=inference_config,
model_dir=MODEL_DIR)
# Get path to saved weights
# Either set a specific path or find last trained weights
# model_path = os.path.join(ROOT_DIR, ".h5 file name here")
model_path = model.find_last()[1]
# Load trained weights (fill in path to trained weights here)
assert model_path != "", "Provide path to trained weights"
print("Loading weights from ", model_path)
model.load_weights(model_path, by_name=True)
# Validation dataset
dataset_val = CellsDataset()
# dataset_val.load_cells(IMAGES_DIR, config.IMAGE_SHAPE[0], config.IMAGE_SHAPE[1])
# dataset_val.load_cells(IMAGES_DIR, config.IMAGE_SHAPE[0], config.IMAGE_SHAPE[1])
dataset_val.load_cells(IMAGES_DIR)
dataset_val.prepare()
# Load and display random samples
image_ids = np.random.choice(dataset_train.image_ids, 1)
print(image_ids)
for image_id in image_ids:
image = dataset_train.load_image(image_id)
mask, class_ids = dataset_train.load_mask(image_id)
visualize.display_top_masks(image, mask, class_ids,
dataset_train.class_names)
# Create model in training mode
model = modellib.MaskRCNN(mode="training",
config=config,
model_dir=MODEL_DIR)
# from keras.utils import plot_model
# plot_model(model, to_file='model_training_mode.png', show_shapes=True)
# Which weights to start with?
init_with = "imagenet" # imagenet, coco, or last
if init_with == "imagenet":
model.load_weights(model.get_imagenet_weights(), by_name=True)
elif init_with == "coco":
# Load weights trained on MS COCO, but skip layers that
# are different due to the different number of classes
# See README for instructions to download the COCO weights
model.load_weights(COCO_MODEL_PATH,