def test_on_folder(model, folder_path, save_path='test_outputs/'):
if not os.path.exists(save_path):
os.mkdir(save_path)
files = glob.glob(folder_path + '/*.jpg')
for i in range(len(files)):
image_id = i
image = skimage.io.imread(files[image_id])
results = model.detect(image[np.newaxis], verbose=0)
results = results[0]
class_names = ['slum'] * (len(results['class_ids']) + 1)
mask = results['masks']
file_to_save = save_path + '/pred_' + str(image_id) + '.jpg'
visualize.save_instances(image,
results['rois'],
results['masks'],
results['class_ids'],
class_names,
file_to_save,
results['scores'],
ax=None,
show_bbox=False,
show_mask=True,
title="Predictions " + str(image_id))
#Uncomment to visualize using matpltolib.
"""
def do_detech_buildings(self):
images_path, ouput_path = self.images_path, self.ouput_path
i = 0
res = []
tif_tans = TIF_TRANS()
for image_name in os.listdir(images_path):
i += 1
img_path = os.path.join(images_path, image_name)
# image = Image.open(imgpath).convert('RGB')
image = skimage.io.imread(img_path)
if len(image.shape) == 2:
image = image[:, :, np.newaxis]
image = np.concatenate((image, image, image), axis=2)
w, h, _ = image.shape # w = 400,h = 400
results = model.detect([image], verbose=1)
# Visualize results
r = results[0]
visualize.save_instances(image,
r['rois'],
r['masks'],
r['class_ids'],
class_names,
r['scores'],
save_name=image_name,
save_path=ouput_path)
def do_detech_roads(self):
images_path, ouput_path = self.images_path, self.ouput_path
i = 0
res = []
tif_tans = TIF_TRANS()
for image_name in os.listdir(images_path):
i += 1
img_path = os.path.join(images_path, image_name)
# image = Image.open(imgpath).convert('RGB')
image = skimage.io.imread(img_path)
if len(image.shape) == 2:
image = image[:, :, np.newaxis]
image = np.concatenate((image, image, image), axis=2)
w, h, _ = image.shape # w = 400,h = 400
results = model.detect([image], verbose=1)
# Visualize results
r = results[0]
visualize.save_instances(image,
r['rois'],
r['masks'],
r['class_ids'],
class_names,
r['scores'],
save_name=image_name,
save_path=ouput_path)
dis, offset_xy = self.center_point(r['rois'], w, h)
m = re.match(r'(\d+)_(\d+)_(\d+)_(\d+).jpg', image_name)
row_point, col_point = int(m.group(3)), int(m.group(4))
x_before, y_before = tif_tans.imagexy2geo(col_point, row_point)
x_after, y_after = tif_tans.imagexy2geo(col_point + offset_xy[1],
row_point + offset_xy[0])
temp = [
offset_xy[0], offset_xy[1], dis, x_before, y_before, x_after,
y_after, img_path
]
res.append(temp)
temp_str = [str(val) for val in temp]
logging.info('_'.join(temp_str))
res_data_frame = pd.DataFrame(res,
columns=[
'offset_x', 'offset_y', 'dis',
'x_before', 'y_before', 'x_after',
'y_after', 'img_path'
])
all_patch_res = res_data_frame[res_data_frame['dis'] != 0]
all_patch_res.to_csv(self.all_patch_res_path)
cluster_res = self.culster(all_patch_res[['offset_x', 'offset_y']])
cluster_res.to_csv(self.culster_csv)
filter_patch_res = all_patch_res[cluster_res['jllable'] == 0]
filter_patch_res.to_csv(self.filter_patch_res_path)
def do_detech_roads(self, image, image_name, res):
ouput_path = self.save_path
img_path = os.path.join(ouput_path, image_name)
w, h, _ = image.shape # w = 400,h = 400
results = self.model.detect([image], verbose=1)
# Visualize results
r = results[0]
visualize.save_instances(image, r['rois'], r['masks'], r['class_ids'],
self.class_names, r['scores'], save_name=image_name, save_path=ouput_path)
dis, offset_xy, is_real = self.center_point(r['rois'], w, h)
m = re.match(r'(\d+)_(\d+)_(\d+).jpg', image_name)
row_point, col_point = int(m.group(2)), int(m.group(3))
x_before, y_before = self.imagexy2geo(col_point, row_point)
x_after, y_after = self.imagexy2geo(col_point + offset_xy[1], row_point + offset_xy[0])
temp = [offset_xy[0], offset_xy[1], dis, x_before, y_before, x_after, y_after, is_real, img_path]
res.append(temp)
temp_str = [str(val) for val in temp]
logging.info('_'.join(temp_str))
def test_on_file(model, img):
image = skimage.io.imread(img)
results = model.detect(image[np.newaxis], verbose=0)
results = results[0]
class_names = ['slum'] * (len(results['class_ids']) + 1)
mask = results['masks']
visualize.save_instances(
image,
results['rois'],
results['masks'],
results['class_ids'],
class_names,
'/media/shaheem/Data/Projects/PycharmProjects/djangoProject/templates/assets/results/pred.jpg',
results['scores'],
ax=None,
show_bbox=False,
show_mask=True,
title="Predictions")
def main(args):
# Get the parameters from args
pb_filepath = args.model_path
InIMAGE_DIR = args.test_images_path
OutIMAGE_DIR = args.test_results_path
# Load the .pb file
tf.reset_default_graph()
with tf.gfile.GFile(pb_filepath, 'rb') as f:
graph_def = tf.GraphDef()
graph_def.ParseFromString(f.read())
_ = tf.import_graph_def(graph_def, name='')
print('Graph loaded.')
image_names = os.listdir(InIMAGE_DIR) # List of images
images = [os.path.join(InIMAGE_DIR, image)
for image in image_names] # List of images with absolute paths
sess = tf.InteractiveSession()
class_names = ["BG", "building"]
for i, image in enumerate(images):
print("Processing:", image_names[i])
start = time.time()
output_path = os.path.join(OutIMAGE_DIR, image_names[i][:-4] + ".png")
# Load image
image = imageio.imread(image)
# remove alpha channel
image = image[:, :, :3]
results = detect(sess, [image], inference_config)
r = results[0]
end = time.time()
print("Processing time = {} seconds \n".format(end - start))
fig = visualize.save_instances(image, r['rois'], r['masks'],
r['class_ids'], class_names,
r['scores'])
plt.imsave(output_path, fig)
print('Done')
def save_detect_list(model, image_dir=None):
image_path = os.listdir(image_dir)
print(image_path[0])
try:
shutil.rmtree('draw')
except:
pass
try:
os.mkdir('draw')
except:
pass
for image_name in image_path:
image = skimage.io.imread(image_dir+image_name)
# Detect objects
predictions = model.detect([image], verbose=1)[0]
scores = predictions['scores']
if len(scores)>0:
index = np.argmax(scores)
box = predictions['rois'][index]
score = scores[index]
visualize.save_instances(image, predictions['rois'], predictions['masks'], predictions['class_ids'],
class_names,image_name, predictions['scores'])
def test_on_img(model, file, id_):
image = skimage.io.imread(file)
if image.shape[-1] == 4:
image = image[..., :3]
results = model.detect(image[np.newaxis], verbose=0)
results = results[0]
class_names = ['slum'] * (len(results['class_ids']) + 1)
mask = results['masks']
file_to_save = 'static/images/pred_' + str(id_) + '.jpg'
_, img = visualize.save_instances(image,
results['rois'],
results['masks'],
results['class_ids'],
class_names,
file_to_save,
results['scores'],
ax=get_ax(0),
show_bbox=False,
show_mask=True,
title="Predictions " + str(id_))
return img, file_to_save
# image_id = random.choice(dataset.image_ids)
image_id = 1473
image, image_meta, gt_class_id, gt_bbox = \
modellib.load_image_gt(dataset, config, image_id)
info = dataset.image_info[image_id]
print("image ID: {}.{} ({}) {}".format(info["source"], info["id"],
image_id,
dataset.image_reference(image_id)))
# Run object detection
results = model.detect([image], verbose=1)
# Display results
ax = get_ax(1)
r = results[0]
save_path = "{}_new_weights".format(image_id)
image = dataset.load_image(image_id)
visualize.save_instances(image,
r['rois'],
gt_bbox,
r['class_ids'],
gt_class_id,
dataset.class_names,
r['scores'],
ax=ax,
title="Predictions_{}".format(info["id"]),
path=save_path,
show_mask=False)
print("gt_class_id", gt_class_id)
print("gt_bbox", gt_bbox)
'keyboard', 'cell phone', 'microwave', 'oven', 'toaster', 'sink',
'refrigerator', 'book', 'clock', 'vase', 'scissors', 'teddy bear',
'hair drier', 'toothbrush'
]
## Run Object Detection
filenames = glob.glob(os.path.join(InIMAGE_DIR, "*.jpg"))
for counter, fl in enumerate(filenames):
print("counter = {:5d}".format(counter))
image_name = fl.split('/')[-1]
output_path = os.path.join(OutIMAGE_DIR, image_name)
if os.path.isfile(output_path):
continue
# image = skimage.io.imread(os.path.join(IMAGE_DIR, random.choice(file_names)))
image = skimage.io.imread(fl)
# remove alpha channel
image = image[:, :, :3]
results = model.detect([image], verbose=1)
r = results[0]
# visualize.display_instances(image, r['rois'], r['masks'], r['class_ids'], class_names, r['scores'])
fig = visualize.save_instances(image, r['rois'], r['masks'],
r['class_ids'], class_names, r['scores'])
# fig.savefig(output_path, bbox_inches='tight', pad_inches=0)
# print(fig.shape)
plt.imsave(output_path, fig)
# fig.imsave(output_path)
uuids = ["c927d51b-92852659"]
# uuids = ["b1d0a191-06deb55d"]
ids = get_ids_from_uuids(dataset, uuids)
# ids = [random.choice(dataset.image_ids)]
# ids = [1536]
for image_id in ids:
image, _, gt_class_id, gt_bbox = modellib.load_image_gt(dataset, config, image_id)
info = dataset.image_info[image_id]
print("image ID: {}.{} ({}) {}".format(info["source"], info["id"], image_id,
dataset.image_reference(image_id)))
# Run object detection
results = model.detect([image], verbose=1, gpi_type=config.GPI_TYPE)
# Display results
ax = get_ax(1)
r = results[0]
# image = dataset.load_image(image_id)
visualize.save_instances(image, r['rois'], gt_bbox, r['class_ids'], gt_class_id, dataset.class_names,
r['scores'],
ax=ax, title="Predictions_{}".format(info["id"]),
path="{}/{}_{}.jpg".format(args.save_path, args.model.split('/')[-2], info["id"]),
show_mask=False)
if r['relation_attention'] is not None:
visualize.draw_attention(r['rois'], r['relation_attention'], image, info["id"])
print("gt_class_id", gt_class_id)
print("gt_bbox", gt_bbox)
print("End Graph Detector Prediction")
class_names = ['BG', 'apple']
# ## Run Object Detection
# In[ ]:
# Load a random image from the images folder
#file_names = next(os.walk(IMAGE_DIR))[2]
#image = skimage.io.imread(os.path.join(IMAGE_DIR, random.choice(file_names)))
for root, dirs, files in os.walk(IMAGE_DIR, topdown=False):
for name in files:
inputFilePath = os.path.join(root, name)
outputFilePath = os.path.join(ROOT_DIR + "/images/AusTestOutput", name)
image = skimage.io.imread(inputFilePath)
# Run detection
results = model.detect([image], verbose=1)
# Visualize results
r = results[0]
#visualize.display_instances(image, r['rois'], r['masks'], r['class_ids'],
# class_names, r['scores'])
visualize.save_instances(image,
r['rois'],
r['masks'],
r['class_ids'],
class_names,
r['scores'],
savepath=outputFilePath)
# In[ ]:
def _get_detections_annotations(dataset,
model,
save_path=None,
config=None,
batch_size=1):
"""
Get the detections from the model using the generator.
The result is a list of lists such that the size is:
all_detections[num_images][num_classes] = detections[num_detections, 4 + num_classes]
# Arguments
generator : The generator used to run images through the model.
model : The model to run on the images.
save_path : The path to save the images with visualized detections to.
# Returns
A list of lists containing the detections for each image in the data set.
"""
all_detections = [[None for i in range(dataset.num_labels())]
for j in range(dataset.size())]
all_annotations = [[None for i in range(dataset.num_labels())]
for j in range(dataset.size())]
image_ids = dataset.image_ids
size = len(image_ids)
t_prediction = 0
t_start = time.time()
# Decide batches per epoch
if size % batch_size == 0:
num_of_batches_per_epoch = size / batch_size
else:
num_of_batches_per_epoch = size / batch_size + 1
for batch in range(num_of_batches_per_epoch):
# Define number of samples per batch
if batch_size * (batch + 1) >= size:
nof_samples_per_batch = size - batch_size * batch
else:
nof_samples_per_batch = batch_size
image_lst = []
gt_class_id_lst = []
gt_bbox_lst = []
for current_index in range(nof_samples_per_batch):
# Get index from files
ind = batch * batch_size + current_index
image_id = image_ids[ind]
# Get data
image, _, gt_class_id, gt_bbox = modellib.load_image_gt(
dataset, config, image_id)
# Append
image_lst.append(image)
gt_class_id_lst.append(gt_class_id)
gt_bbox_lst.append(gt_bbox)
# Run detection
t = time.time()
r_lst = model.detect(image_lst, verbose=0, gpi_type=config.GPI_TYPE)
t_prediction += (time.time() - t)
for current_index in range(nof_samples_per_batch):
# Get index from files
i = batch * batch_size + current_index
# Get data
image_id = image_ids[i]
gt_class_id = gt_class_id_lst[current_index]
gt_bbox = gt_bbox_lst[current_index]
r = r_lst[current_index]
image_boxes = r["rois"]
image_labels = r["class_ids"]
image_scores = r["scores"]
id = dataset.image_info[image_id]['id']
if save_path is not None:
image = dataset.load_image(image_id)
visualize.save_instances(image,
r['rois'],
gt_bbox,
r['class_ids'],
gt_class_id,
dataset.class_names,
r['scores'],
ax=None,
show_mask=False,
path=os.path.join(
save_path, "{}.jpg".format(id)),
title="Predictions_{}".format(id))
# select detections - [[num_boxes, y1, x1, y2, x2, score, class_id]]
image_detections = np.concatenate([
image_boxes,
np.expand_dims(image_scores, axis=1),
np.expand_dims(image_labels, axis=1)
],
axis=1)
# load the annotations - [[num_boxes, y1, x1, y2, x2, class_id]]
annotations = np.concatenate(
[gt_bbox, np.expand_dims(gt_class_id, axis=1)], axis=1)
# copy detections to all_detections
for label in range(dataset.num_labels()):
all_detections[i][label] = image_detections[
image_detections[:, -1] == label, :-1]
# copy detections to all_annotations
for label in range(dataset.num_labels()):
all_annotations[i][label] = annotations[annotations[:, 4] ==
label, :4].copy()
# print('{}/{}'.format(i + 1, dataset.size()))
print('Batch {}/{}'.format(batch + 1, num_of_batches_per_epoch))
print("Prediction time: {}. Average {}/image".format(
t_prediction, t_prediction / len(image_ids)))
print("Total time: ", time.time() - t_start)
return all_detections, all_annotations
def main():
import argparse
# Parse command line arguments
parser = argparse.ArgumentParser(
description='Train Mask R-CNN on satellite images.')
parser.add_argument("--size",
metavar="size",
default="small",
help="'large' or 'small' images")
args = parser.parse_args()
print("Image Type: ", args.size)
# Root directory of the project
ROOT_DIR = os.path.abspath("../")
# Import Mask RCNN
sys.path.append(ROOT_DIR) # To find local version of the library
from mrcnn import utils
import mrcnn.model as modellib
from mrcnn import visualize
# from mrcnn.visualize import save_image # added by JX
# Import COCO config
sys.path.append(os.path.join(ROOT_DIR, "samples/coco/")) # To find local version
import sate
#matplotlib inline
# Directory to save logs and trained model
MODEL_DIR = os.path.join(ROOT_DIR, "logs")
# Local path to trained weights file
COCO_MODEL_PATH = os.path.join(ROOT_DIR, "mask_rcnn_sate_b2_0071.h5")
#COCO_MODEL_PATH = os.path.join(ROOT_DIR, "mask_rcnn_sate_0055.h5")
#COCO_MODEL_PATH = os.path.join(ROOT_DIR, "crowdai_pretrained_weights.h5")
# # Download COCO trained weights from Releases if needed
# if not os.path.exists(COCO_MODEL_PATH):
# utils.download_trained_weights(COCO_MODEL_PATH)
# Directory of images to run detection on
# IMAGE_DIR = os.path.join(ROOT_DIR, "images")
InIMAGE_DIR = "/home/ashwin/Desktop/Test/input"
OutIMAGE_DIR = "/home/ashwin/Desktop/Test/output"
if not os.path.isdir(OutIMAGE_DIR):
os.makedirs(OutIMAGE_DIR)
if args.size == 'large':
Image.MAX_IMAGE_PIXELS = 1600 * 1600 * 10 * 10
img_name = os.listdir(InIMAGE_DIR)[0]
img = os.path.join(InIMAGE_DIR, img_name)
num_tiles = 64
print("Slicing image into {} slices".format(num_tiles))
tiles = image_slicer.slice(img, num_tiles)
## Configurations
class InferenceConfig(sate.CocoConfig):
# 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 and Load Trained Weights
# Create model object in inference mode.
model = modellib.MaskRCNN(mode="inference", model_dir=MODEL_DIR, config=config)
model_json = model.KM.to_json()
with open("model.json", "w") as json_file:
json_file.write(model_json)
print("Loading weights from:",COCO_MODEL_PATH)
# Load weights trained on MS-COCO
model.load_weights(COCO_MODEL_PATH, by_name=True)#, exclude=["mrcnn_bbox_fc", "mrcnn_class_logits", "mrcnn_mask"])
## Class Names
# COCO Class names
# Index of the class in the list is its ID. For example, to get ID of
# the teddy bear class, use: class_names.index('teddy bear')
# class_names = ['Plane', 'Ships', 'Running_Track', 'Helicopter', 'Vehicles', 'Storage_Tanks', 'Tennis_Court',
# 'Basketball_Court', 'Bridge', 'Roundabout', 'Soccer_Field', 'Swimming_Pool', 'baseball_diamond',
# 'Buildings', 'Road', 'Tree', 'People', 'Hangar', 'Parking_Lot', 'Airport', 'Motorcycles', 'Flag',
# 'Sports_Stadium', 'Rail_(for_train)', 'Satellite_Dish', 'Port', 'Telephone_Pole',
# 'Intersection/Crossroads', 'Shipping_Container_Lot', 'Pier', 'Crane', 'Train', 'Tanks', 'Comms_Tower',
# 'Cricket_Pitch', 'Submarine', 'Radar', 'Horse_Track', 'Hovercraft', 'Missiles', 'Artillery',
# 'Racing_Track', 'Vehicle_Sheds', 'Fire_Station', 'Power_Station', 'Refinery', 'Mosques', 'Helipads',
# 'Shipping_Containers', 'Runway', 'Prison', 'Market/Bazaar', 'Police_Station', 'Quarry', 'School',
# 'Graveyard', 'Well', 'Rifle_Range', 'Farm', 'Train_Station', 'Crossing_Point', 'Telephone_Line',
# 'Vehicle_Control_Point', 'Warehouse', 'Body_Of_water', 'Hospital', 'Playground', 'Solar_Panel']
class_names = ['BG','building']
# model_filename = "MRCNN_spacemodel.pkl"
# pickle.dump(model,open(model_filename, 'wb'))
if args.size == 'large':
for tile in tiles:
image = skimage.io.imread(tile.filename)
# remove alpha channel
image = image[:, :, :3]
results = model.detect([image], verbose=1)
r = results[0]
fig = visualize.save_instances(image, r['rois'], r['masks'], r['class_ids'], class_names, r['scores'])
plt.imsave(tile.filename,fig)
tile.image = Image.open(tile.filename)
image = join(tiles)
image.save(os.path.join(OutIMAGE_DIR,img_name))
if args.size == 'small':
gt_flag = 0
# TODO: If any tif files are present, convert them to jpg
## Run Object Detection
filenames = glob.glob(os.path.join(InIMAGE_DIR, "*.jpg"))
with open('./coco/annotations_batch1/train.json') as f: # Not needed for inference
gj = json.load(f)
for counter, fl in enumerate(filenames):
print("counter = {:5d}".format(counter))
image_name = fl.split('/')[-1]
output_path = os.path.join(OutIMAGE_DIR, image_name)
if os.path.isfile(output_path):
continue
# image = skimage.io.imread(os.path.join(IMAGE_DIR, random.choice(file_names)))
image = skimage.io.imread(fl)
# remove alpha channel
image = image[:, :, :3]
results = model.detect([image], verbose=1)
r = results[0]
# visualize.display_instances(image, r['rois'], r['masks'], r['class_ids'], class_names, r['scores'])
#####################################################################################################
if gt_flag == 1:
bbox_actual = []
print("Image ID:",image_name[:-4])
for ind in gj['annotations']:
if ind['image_id'] == image_name[:-4]:
bbox_actual.append(ind['bbox'])
#print("Number of proposals =",len(r['rois']))
#print("Number of ground truth labels =",len(bbox_actual))
#print("Proposal1= ",r['rois'][0])
#print("ActualBBox1= ",bbox_actual[0])
# Ideally it should be a square matrix with high IOU values in the diagonal
#iou_mat = bbox_iou(r['rois'],bbox_actual,bb_format="XYWH")
#print(iou_mat)
#plt.figure()
#plt.imshow(iou_mat, cmap="hot",interpolation="nearest")
#plt.show()
#####################################################################################################
fig = visualize.save_instances(image, r['rois'], r['masks'], r['class_ids'], class_names, r['scores'])
# fig.savefig(output_path, bbox_inches='tight', pad_inches=0)
print("Number of detected buildings =",len(r['rois']))
if gt_flag == 1:
print("Number of ground truth buildings",len(bbox_actual))
# print(fig.shape)
plt.imsave(output_path, fig)
def infer(args):
from mrcnn import model as mrcnn_lib
infer_path = args.path
output = args.output
do_pictures = not args.no_pictures
do_contours = not args.no_contours
model_path = args.model
should_save_masks = not args.no_masks
gt_adapter = args.gt
task_id = args.task
compare_to_gt = gt_adapter is not None
# Retrieve images
images = list(generate_images(infer_path))
# Retrieve gt masks
if compare_to_gt:
adapter_class = locate(gt_adapter + '.' + gt_adapter)
gt_path = infer_path
gt_annotation_map = generate_annotation_map(adapter_class, gt_path,
images, task_id)
# Retrieve model path
model_path = prompt_model(model_path)
# Load model
inference_config = get_inference_config(LeafSegmentorConfig)
if not os.path.exists(output):
os.makedirs(output, exist_ok=True)
model = mrcnn_lib.MaskRCNN(mode="inference",
config=inference_config,
model_dir=output)
model.load_weights(model_path, by_name=True)
model.set_log_dir()
output_dir = model.log_dir
os.makedirs(output_dir, exist_ok=True)
# Infer
inference_dict = {}
IoU_dict = {}
for image_path in tqdm(images):
inference_dict[image_path] = []
image_name = os.path.basename(image_path)
image = np.array(Image.open(image_path))
r = model.detect([image])[0]
if should_save_masks:
save_masks(r['masks'], output_dir, image_name)
if do_pictures:
output_file_path = os.path.join(output_dir, image_name)
_, ax = plt.subplots(1, figsize=(16, 16))
visualize.save_instances(image,
r['rois'],
r['masks'],
r['class_ids'], ['BG', 'leave'],
r['scores'],
save_to=output_file_path,
ax=ax)
if do_contours:
inference_dict[image_path], txt_contours = get_contours(r)
for i, leaf_contour in enumerate(txt_contours):
for j, polygon_contour in enumerate(leaf_contour):
contour_file_name = os.path.join(output_dir, os.path.splitext(image_name)[0]) + \
"_" + str(i).zfill(3) + "_" + str(j) + ".txt"
np.savetxt(contour_file_name,
polygon_contour,
fmt='%.1f',
delimiter=' , ')
if compare_to_gt:
gt_masks = get_all_masks(gt_annotation_map[image_path], image_path)
gt_image_name = ".".join(image_name.split(".")[:-1]) + "_GT.png"
save_masks(gt_masks, output_dir, gt_image_name)
IoU_dict[image_path] = calculate_iou(image_name, r['masks'],
gt_masks)
if do_contours:
with open(os.path.join(output_dir, CONTOUR_FILE_NAME), 'w') as f:
f.write(json.dumps(inference_dict, indent=2))
if compare_to_gt:
total_score = sum(IoU_dict.values()) / len(IoU_dict)
print("average IoU scores: " + str(total_score))
