def test_image(image_info, disp_maps_dir, disp_map_count, disp_max_abs_value,
batch_size, ds_fac_list, run_name_list,
model_disp_max_abs_value, thresholds, test_output_dir):
# --- Load data --- #
ori_image, ori_metadata, ori_gt_polygons = read.load_gt_data(
config_test.DATASET_RAW_DIR, image_info["city"], image_info["number"])
image_name = read.IMAGE_NAME_FORMAT.format(city=image_info["city"],
number=image_info["number"])
# --- Load disp maps --- #
disp_maps = load_disp_maps(disp_maps_dir, image_info, disp_map_count)
test.test_image_with_gt_polygons_and_disp_maps(
image_name,
ori_image,
ori_metadata,
ori_gt_polygons,
disp_maps,
disp_max_abs_value,
batch_size,
ds_fac_list,
run_name_list,
model_disp_max_abs_value,
thresholds,
test_output_dir,
output_shapefiles=config_test.OUTPUT_SHAPEFILES)
def test_image(runs_dirpath, dataset_raw_dirpath, image_info, batch_size, ds_fac_list, run_name_list,
model_disp_max_abs_value, output_dir, output_shapefiles):
# --- Load data --- #
# CHANGE the arguments of the load_gt_data() function if using your own and it does not take the same arguments:
ori_image, ori_metadata, gt_polygons = read_bradbury_buildings.load_gt_data(dataset_raw_dirpath, image_info["city"],
image_info["number"])
if INPUT_POLYGONS_FILENAME_EXTENSION is not None:
gt_polygons = read_bradbury_buildings.load_polygons(dataset_raw_dirpath, image_info["city"], image_info["number"], INPUT_POLYGONS_FILENAME_EXTENSION)
else:
gt_polygons = gt_polygons
if gt_polygons is not None:
# CHANGE the arguments of the IMAGE_NAME_FORMAT format string if using your own and it does not take the same arguments:
image_name = read_bradbury_buildings.IMAGE_NAME_FORMAT.format(city=image_info["city"], number=image_info["number"])
print_utils.print_info("Processing image {}".format(image_name))
aligned_gt_polygons = test.test_align_gt(runs_dirpath, ori_image, ori_metadata, gt_polygons, batch_size,
ds_fac_list, run_name_list,
model_disp_max_abs_value, output_dir, image_name,
output_shapefiles=output_shapefiles)
# Save aligned_gt_polygons in dataset dir:
aligned_gt_polygons_filepath = read_bradbury_buildings.get_polygons_filepath(dataset_raw_dirpath, image_info["city"], image_info["number"], ALIGNED_GT_POLYGONS_FILENAME_EXTENSION)
os.makedirs(os.path.dirname(aligned_gt_polygons_filepath), exist_ok=True)
np.save(aligned_gt_polygons_filepath, aligned_gt_polygons)
def test_detect_new_buildings(image_info, batch_size, ds_fac_list,
run_name_list, model_disp_max_abs_value,
thresholds, test_output_dir):
# --- Load data --- #
ori_image, ori_metadata, ori_gt_polygons = read.load_gt_data(
config_test.DATASET_RAW_DIR, image_info["city"], image_info["number"])
image_name = read.IMAGE_NAME_FORMAT.format(city=image_info["city"],
number=image_info["number"])
polygonization_params = {
"fill_threshold": FILL_THRESHOLD,
"outline_threshold": OUTLINE_THRESHOLD,
"selem_width": SELEM_WIDTH,
"iterations": ITERATIONS,
}
test.test_detect_new_buildings(
image_name,
ori_image,
ori_metadata,
ori_gt_polygons,
batch_size,
ds_fac_list,
run_name_list,
model_disp_max_abs_value,
polygonization_params,
thresholds,
test_output_dir,
output_shapefiles=config_test.OUTPUT_SHAPEFILES)
def test_image(image_info, batch_size, ds_fac_list, run_name_list, model_disp_max_abs_value, thresholds, test_output_dir):
# --- Load data --- #
# CHANGE the arguments of the load_gt_data() function if using your own and it does not take the same arguments:
ori_image, ori_metadata, ori_disp_polygons = read.load_gt_data(config_test.DATASET_RAW_DIR, image_info["city"], image_info["number"])
# CHANGE the arguments of the IMAGE_NAME_FORMAT format string if using your own and it does not take the same arguments:
image_name = read.IMAGE_NAME_FORMAT.format(city=image_info["city"], number=image_info["number"])
ori_gt_polygons = []
test.test(ori_image, ori_metadata, ori_gt_polygons, ori_disp_polygons, batch_size, ds_fac_list, run_name_list,
model_disp_max_abs_value, thresholds, test_output_dir, image_name, output_shapefiles=config_test.OUTPUT_SHAPEFILES)
def test_image(runs_dirpath, dataset_raw_dirpath, image_info, disp_maps_dir, disp_map_count, disp_max_abs_value,
batch_size, ds_fac_list, run_name_list,
model_disp_max_abs_value, thresholds, test_output_dir, output_shapefiles):
# --- Load data --- #
ori_image, ori_metadata, ori_gt_polygons = read_inria.load_gt_data(dataset_raw_dirpath, image_info["city"],
image_info["number"])
image_name = read_inria.IMAGE_NAME_FORMAT.format(city=image_info["city"], number=image_info["number"])
# --- Load disp maps --- #
disp_maps = load_disp_maps(disp_maps_dir, image_info, disp_map_count)
test.test_image_with_gt_polygons_and_disp_maps(runs_dirpath, image_name, ori_image, ori_metadata, ori_gt_polygons,
disp_maps,
disp_max_abs_value, batch_size, ds_fac_list, run_name_list,
model_disp_max_abs_value, thresholds, test_output_dir,
output_shapefiles=output_shapefiles)
def load_data(raw_dirpath, patch_info_list):
image_list = []
polygons_list = []
for patch_info in patch_info_list:
additional_args = {
"overwrite_polygon_dir_name": POLYGON_DIRNAME,
}
image, metadata, polygons = read.load_gt_data(
raw_dirpath,
patch_info["city"],
patch_info["number"],
additional_args=additional_args)
scaled_bbox = np.array(patch_info["bbox"] / patch_info["scale_factor"],
dtype=np.int)
p_image = image[scaled_bbox[0]:scaled_bbox[2],
scaled_bbox[1]:scaled_bbox[3], :]
image_list.append(p_image)
p_polygons = polygon_utils.crop_polygons_to_patch_if_touch(
polygons, scaled_bbox)
polygons_list.append(p_polygons)
return image_list, polygons_list
def test_image(runs_dirpath, dataset_raw_dirpath, image_info, disp_maps_dir, disp_map_count, disp_max_abs_value,
batch_size, ds_fac_list, run_name_list, model_disp_max_abs_value, thresholds, test_output_dir,
output_shapefiles):
# --- Load data --- #
ori_image, ori_metadata, ori_gt_polygons = read_bradbury_buildings.load_gt_data(dataset_raw_dirpath,
image_info["city"],
image_info["number"])
image_name = read_bradbury_buildings.IMAGE_NAME_FORMAT.format(city=image_info["city"], number=image_info["number"])
# --- Randomly drop some polygons --- #
if KEEP_PROB < 1:
ori_gt_polygons = drop_items(ori_gt_polygons, KEEP_PROB)
# --- Load disp maps --- #
disp_maps = load_disp_maps(disp_maps_dir, image_info, disp_map_count)
test.test_image_with_gt_polygons_and_disp_maps(runs_dirpath, image_name, ori_image, ori_metadata, ori_gt_polygons,
disp_maps,
disp_max_abs_value, batch_size, ds_fac_list, run_name_list,
model_disp_max_abs_value, thresholds, test_output_dir,
output_shapefiles=output_shapefiles)
def generate_disp_maps(dataset_raw_dir, image_info, disp_map_params,
thresholds, output_dir):
print("Generating {} displacement maps for {} {}...".format(
disp_map_params["disp_map_count"], image_info["city"],
image_info["number"]))
disp_map_filename_format = "{}.disp_{:02d}.disp_map.npy"
accuracies_filename_format = "{}.disp_{:02d}.accuracy.npy"
# --- Load data --- #
ori_image, ori_metadata, ori_gt_polygons = read.load_gt_data(
dataset_raw_dir, image_info["city"], image_info["number"])
image_name = read.IMAGE_NAME_FORMAT.format(city=image_info["city"],
number=image_info["number"])
spatial_shape = ori_image.shape[:2]
ori_normed_disp_field_maps = math_utils.create_displacement_field_maps(
spatial_shape, disp_map_params["disp_map_count"],
disp_map_params["disp_modes"], disp_map_params["disp_gauss_mu_range"],
disp_map_params["disp_gauss_sig_scaling"])
disp_polygons_list = test.generate_disp_data(
ori_normed_disp_field_maps, ori_gt_polygons,
disp_map_params["disp_max_abs_value"])
# Save disp maps and accuracies individually
for i, (ori_normed_disp_field_map, disp_polygons) in enumerate(
zip(ori_normed_disp_field_maps, disp_polygons_list)):
disp_map_filename = disp_map_filename_format.format(image_name, i)
disp_map_filepath = os.path.join(output_dir, disp_map_filename)
np.save(disp_map_filepath, ori_normed_disp_field_map)
accuracies_filename = accuracies_filename_format.format(image_name, i)
accuracies_filepath = os.path.join(output_dir, accuracies_filename)
integer_thresholds = [
threshold for threshold in thresholds
if (int(threshold) == threshold)
]
accuracies = test.measure_accuracies(ori_gt_polygons, disp_polygons,
integer_thresholds,
accuracies_filepath)
def compute_grads(raw_dirpath, runs_dirpath, run_name, ds_fac,
overwrite_config, tile_info_list, polygon_dirname,
output_dirname, output_filepath_format):
# -- Params:
# Setup run dir and load config file
run_dir = run_utils.setup_run_dir(runs_dirpath, run_name)
_, checkpoints_dir = run_utils.setup_run_subdirs(run_dir)
config = run_utils.load_config(config_dirpath=run_dir)
# --- Instantiate model
output_res = model.MapAlignModel.get_output_res(
overwrite_config["input_res"], config["pool_count"])
map_align_model = model.MapAlignModel(
config["model_name"], overwrite_config["input_res"],
config["add_image_input"], config["image_channel_count"],
config["image_feature_base_count"], config["add_poly_map_input"],
config["poly_map_channel_count"],
config["poly_map_feature_base_count"],
config["common_feature_base_count"], config["pool_count"],
config["add_disp_output"], config["disp_channel_count"],
config["add_seg_output"], config["seg_channel_count"], output_res,
overwrite_config["batch_size"], config["loss_params"],
config["level_loss_coefs_params"], config["learning_rate_params"],
config["weight_decay"], config["image_dynamic_range"],
config["disp_map_dynamic_range_fac"], config["disp_max_abs_value"])
map_align_model.setup_compute_grads() # Add ops to compute gradients
saver = tf.train.Saver(save_relative_paths=True)
with tf.Session() as sess:
# Restore checkpoint
restore_checkpoint_success = map_align_model.restore_checkpoint(
sess, saver, checkpoints_dir)
if not restore_checkpoint_success:
sys.exit('No checkpoint found in {}'.format(checkpoints_dir))
# Compute patch count
patch_total_count = 0
for tile_info in tile_info_list:
patch_total_count += len(tile_info["bbox_list"])
pbar = tqdm(total=patch_total_count,
desc="Computing patch gradients: ")
for tile_info in tile_info_list:
# --- Path setup:
unused_filepath = output_filepath_format.format(
dir=raw_dirpath,
fold=tile_info["fold"],
out_dir=output_dirname,
tile="",
b0=0,
b1=0,
b2=0,
b3=0,
out_name="",
ext="")
os.makedirs(os.path.dirname(unused_filepath), exist_ok=True)
tile_name = read.IMAGE_NAME_FORMAT.format(
city=tile_info["city"], number=tile_info["number"])
# Compute grads for that image
additional_args = {
"overwrite_polygon_dir_name": polygon_dirname,
}
# t = time.clock()
image, metadata, polygons = read.load_gt_data(
raw_dirpath,
tile_info["city"],
tile_info["number"],
additional_args=additional_args)
# t_read = time.clock() - t
# Downsample
image, polygons = process_utils.downsample_data(
image, metadata, polygons, ds_fac,
config["reference_pixel_size"])
spatial_shape = image.shape[:2]
# Draw polygon map
# t = time.clock()
polygon_map = polygon_utils.draw_polygon_map(polygons,
spatial_shape,
fill=True,
edges=True,
vertices=True)
# t_draw = time.clock() - t
t_grads = 0
t_save = 0
for bbox in tile_info["bbox_list"]:
p_im = image[bbox[0]:bbox[2], bbox[1]:bbox[3], :]
p_polygon_map = polygon_map[bbox[0]:bbox[2],
bbox[1]:bbox[3], :]
# p_polygons = polygon_utils.crop_polygons_to_patch_if_touch(polygons, bbox)
# Grad compute
t = time.clock()
grads = map_align_model.compute_grads(sess, p_im,
p_polygon_map)
t_grads += time.clock() - t
# Saving
t = time.clock()
flattened_grads_x = get_flattened_gradients(grads["x"])
flattened_grads_y = get_flattened_gradients(grads["y"])
flattened_grads = np.stack(
[flattened_grads_x, flattened_grads_y], axis=-1)
# # Save patch for later visualization
# im_filepath = output_filepath_format.format(dir=raw_dirpath, fold=tile_info["fold"],
# out_dir=output_dirname, tile=tile_name,
# b0=bbox[0], b1=bbox[1], b2=bbox[2], b3=bbox[3],
# out_name="image", ext="png")
# skimage.io.imsave(im_filepath, p_im)
# # Save polygons as well
# polygons_filepath = output_filepath_format.format(dir=raw_dirpath, fold=tile_info["fold"],
# out_dir=output_dirname, tile=tile_name,
# b0=bbox[0], b1=bbox[1], b2=bbox[2], b3=bbox[3],
# out_name="polygons", ext="npy")
# np.save(polygons_filepath, p_polygons)
# Save grads
grads_filepath = output_filepath_format.format(
dir=raw_dirpath,
fold=tile_info["fold"],
out_dir=output_dirname,
tile=tile_name,
b0=bbox[0],
b1=bbox[1],
b2=bbox[2],
b3=bbox[3],
out_name="grads",
ext="npy")
np.save(grads_filepath, flattened_grads)
t_save += time.clock() - t
pbar.update(len(tile_info["bbox_list"]))
pbar.set_postfix(t_grads=t_grads, t_save=t_save)
pbar.close()
def process_image(dataset_raw_dirpath, image_info, patch_stride, patch_res, downsampling_factors, disp_max_abs_value,
include_polygons,
downsampling_factor_writers):
"""
Writes to all the writers (one for each resolution) all sample patches extracted from the image_info.
:param raw_dirpath:
:param image_info:
:param patch_stride:
:param patch_res:
:param downsampling_factors:
:param disp_max_abs_value:
:param include_polygons:
:param downsampling_factor_writers:
:return:
"""
ori_image, ori_metadata, ori_gt_polygons = read.load_gt_data(dataset_raw_dirpath, image_info["city"],
image_info["number"])
if ori_gt_polygons is None:
return False
ori_gt_polygons = polygon_utils.polygons_remove_holes(ori_gt_polygons) # TODO: Remove
# Remove redundant vertices
ori_gt_polygons = polygon_utils.simplify_polygons(ori_gt_polygons, tolerance=1)
# visualization.init_figures(["gt_data"], figsize=(60, 40))
# visualization.plot_example_polygons("gt_data", ori_image, ori_gt_polygons)
# Create displacement maps
ori_normed_disp_field_maps = math_utils.create_displacement_field_maps(ori_image.shape[:2], config.DISP_MAP_COUNT,
config.DISP_MODES,
config.DISP_GAUSS_MU_RANGE,
config.DISP_GAUSS_SIG_SCALING) # TODO: uncomment
# ori_normed_disp_field_maps = np.zeros((config.DISP_MAP_COUNT, ori_image.shape[0], ori_image.shape[1], 2)) # TODO: remove
# # TODO: remove
# np.random.seed(seed=0)
# colors = np.random.randint(0, 255, size=(len(downsampling_factors), 3), dtype=np.uint8)
for index, downsampling_factor in enumerate(downsampling_factors):
print("downsampling_factor: {}".format(downsampling_factor))
# Downsample ground-truth
image, gt_polygons, normed_disp_field_maps = downsample_gt_data(ori_image, ori_metadata, ori_gt_polygons,
ori_normed_disp_field_maps, downsampling_factor)
spatial_shape = image.shape[:2]
# Random color
# image = np.tile(colors[index], reps=[image.shape[0], image.shape[1], 1]) # TODO: remove
# Draw gt polygon map
gt_polygon_map = polygon_utils.draw_polygon_map(gt_polygons, spatial_shape, fill=True, edges=True,
vertices=True)
# Generate final displacement
disp_polygons_list, disp_polygon_maps = generate_disp_data(normed_disp_field_maps, gt_polygons,
disp_max_abs_value, spatial_shape)
# Compress data
gt_polygons = [polygon.astype(np.float16) for polygon in gt_polygons]
disp_polygons_list = [[polygon.astype(np.float16) for polygon in polygons] for polygons in disp_polygons_list]
disp_field_maps = normed_disp_field_maps * 32767 # int16 max value = 32767
disp_field_maps = np.round(disp_field_maps)
disp_field_maps = disp_field_maps.astype(np.int16)
# Cut sample into patches
if include_polygons:
patches = process_sample_into_patches(patch_stride, patch_res, image, gt_polygon_map,
disp_field_maps, disp_polygon_maps,
gt_polygons, disp_polygons_list)
else:
patches = process_sample_into_patches(patch_stride, patch_res, image, gt_polygon_map,
disp_field_maps, disp_polygon_maps)
for patch in patches:
save_patch_to_tfrecord(patch, downsampling_factor_writers[downsampling_factor])
return True
