def get_true_boxes(foldername,
filename,
smart_resized_target=None,
delim=' ',
limit=1000000):
with open(filename) as f:
lines = f.readlines()[1:limit]
filenames_dict = sorted(set([line.split(delim)[0] for line in lines]))
true_boxes = []
print('\nGet %d true boxes\n' % len(filenames_dict))
bar = progressbar.ProgressBar(max_value=len(filenames_dict))
for b, filename in enumerate(filenames_dict):
bar.update(b)
if not os.path.isfile(foldername + filename): continue
image = Image.open(foldername + filename)
if image is None: continue
width, height = image.size
local_boxes = []
for line in lines:
split = line.split(delim)
if split[0] == filename:
class_ID = int(split[5])
x_min, y_min, x_max, y_max = numpy.array(split[1:5]).astype(
numpy.float)
if smart_resized_target is not None:
x_min, y_min = tools_image.smart_resize_point(
x_min, y_min, width, height, smart_resized_target[1],
smart_resized_target[0])
x_max, y_max = tools_image.smart_resize_point(
x_max, y_max, width, height, smart_resized_target[1],
smart_resized_target[0])
local_boxes.append([x_min, y_min, x_max, y_max, class_ID])
true_boxes.append(local_boxes)
return true_boxes
def get_true_boxes(foldername,
filename,
delim='\t',
smart_resized_target=None,
limit=100):
with open(filename) as f:
lines = f.readlines()[1:limit]
list_filenames = [line.split(' ')[0] for line in lines]
filenames_dict = sorted(set(list_filenames))
true_boxes = []
for filename in filenames_dict:
local_boxes = []
for line in lines:
split = line.split(delim)
if split[0] == filename:
class_ID = int(split[5])
x_min, y_min, x_max, y_max = numpy.array(split[1:5]).astype(
numpy.float)
if smart_resized_target is not None:
image = cv2.imread(foldername + filename)
x_min, y_min = tools_image.smart_resize_point(
x_min, y_min, image.shape[1], image.shape[0],
smart_resized_target[1], smart_resized_target[0])
x_max, y_max = tools_image.smart_resize_point(
x_max, y_max, image.shape[1], image.shape[0],
smart_resized_target[1], smart_resized_target[0])
local_boxes.append([x_min, y_min, x_max, y_max, class_ID])
true_boxes.append(local_boxes)
return true_boxes
def calc_hits_stats_iou(lines_true,
lines_pred,
class_ID,
delim,
folder_annotation,
iuo_th=0.5,
ovp_th=0.5,
ovd_th=0.5):
file_true, file_pred = [], []
coord_true, coord_pred, = [], []
conf_true, conf_pred = [], []
hit_true, hit_pred = [], []
for line in lines_true:
split = line.split(delim)
if int(split[5]) == class_ID:
x_min, y_min, x_max, y_max = int(split[1]), int(split[2]), int(
split[3]), int(split[4])
if not os.path.isfile(folder_annotation + split[0]): continue
image = Image.open(folder_annotation + split[0])
if image is None: continue
width, height = image.size
x_min, y_min = tools_image.smart_resize_point(
x_min, y_min, width, height, 416, 416)
x_max, y_max = tools_image.smart_resize_point(
x_max, y_max, width, height, 416, 416)
file_true.append(split[0])
coord_true.append([x_min, y_min, x_max, y_max])
conf_true.append(float(-1))
hit_true.append(0)
for line in lines_pred:
split = line.split(delim)
if int(split[5]) == class_ID:
x_min, y_min, x_max, y_max = int(split[1]), int(split[2]), int(
split[3]), int(split[4])
if not os.path.isfile(folder_annotation + split[0]): continue
image = Image.open(folder_annotation + split[0])
if image is None: continue
width, height = image.size
x_min, y_min = tools_image.smart_resize_point(
x_min, y_min, width, height, 416, 416)
x_max, y_max = tools_image.smart_resize_point(
x_max, y_max, width, height, 416, 416)
file_pred.append(split[0])
coord_pred.append([x_min, y_min, x_max, y_max])
conf_pred.append(float(split[6]))
hit_pred.append(0)
if iuo_th is not None:
for j, filename_true in enumerate(file_true):
best_i, best_iou, best_conf = None, -1, None
for i, filename_pred in enumerate(file_pred):
if filename_true == filename_pred:
iuo_value = iou(coord_true[j], coord_pred[i])
if iuo_value >= iuo_th and iuo_value > best_iou:
best_i, best_iou, best_conf = i, iuo_value, conf_pred[
i]
if best_i is not None:
hit_true[j], conf_true[j] = 1, best_conf
hit_pred[best_i], conf_pred[best_i] = 1, best_conf
else:
hit_true[j], conf_true[j] = 0, float(-1)
else:
for j, filename_true in enumerate(file_true):
for i, filename_pred in enumerate(file_pred):
if filename_true == filename_pred:
ovp_value, ovd_value = ovelraps(coord_true[j],
coord_pred[i])
if ovp_value >= ovp_th and ovd_value <= ovd_th:
hit_pred[i] = 1
hit_true[j] = 1
conf_true[j] = max(conf_true[j], conf_pred[i])
conf_true, conf_pred = numpy.array(conf_true), numpy.array(conf_pred)
hit_true, hit_pred = numpy.array(hit_true), numpy.array(hit_pred)
file_true, file_pred = numpy.array(file_true), numpy.array(file_pred)
coord_true, coord_pred = numpy.array(coord_true), numpy.array(coord_pred)
return file_true, file_pred, coord_true, coord_pred, conf_true, conf_pred, hit_true, hit_pred
def draw_annotation_boxes(file_annotations,
file_classes,
file_metadata,
path_out,
delim=' '):
tools_IO.remove_files(path_out, create=True)
input_image_size, class_names, anchors, anchor_mask, obj_threshold, nms_threshold = load_metadata(
file_metadata)
mat = tools_IO.load_mat(file_classes, numpy.str)
if len(mat) <= len(class_names):
class_names[:len(mat)] = mat
foldername = '/'.join(file_annotations.split('/')[:-1]) + '/'
with open(file_annotations) as f:
lines = f.readlines()[1:]
boxes_xyxy = numpy.array([line.split(delim)[1:5] for line in lines],
dtype=numpy.int)
filenames = numpy.array([line.split(delim)[0] for line in lines])
class_IDs = numpy.array([line.split(delim)[5] for line in lines],
dtype=numpy.int)
colors = generate_colors(numpy.max(class_IDs) + 1)
true_boxes = get_true_boxes(foldername,
file_annotations, (416, 416),
delim=' ')
if len(true_boxes) > 6:
anchors = annotation_boxes_to_ancors(true_boxes, 6)
descript_ion = []
for filename in set(filenames):
image = cv2.imread(foldername + filename)
image = tools_image.desaturate(image, 0.9)
idx = numpy.where(filenames == filename)
boxes_resized = []
for box in boxes_xyxy[idx]:
x_min, y_min = tools_image.smart_resize_point(
box[0], box[1], image.shape[1], image.shape[0], 416, 416)
x_max, y_max = tools_image.smart_resize_point(
box[2], box[3], image.shape[1], image.shape[0], 416, 416)
boxes_resized.append([x_min, y_min, x_max, y_max, 0])
statuses = are_boxes_preprocessed_well(boxes_resized, anchors,
anchor_mask, len(class_names))
descript_ion.append([filename, 1 * (statuses.sum() == len(statuses))])
for box, class_ID, status in zip(boxes_xyxy[idx], class_IDs[idx],
statuses):
w = 2 if status > 0 else -1
cv2.rectangle(image, (box[0], box[1]), (box[2], box[3]),
colors[class_ID],
thickness=w)
cv2.putText(image, '{0:d} {1:s}'.format(class_ID,
class_names[class_ID]),
(box[0], box[1] - 4), cv2.FONT_HERSHEY_SIMPLEX, 0.6,
colors[class_ID], 1, cv2.LINE_AA)
cv2.imwrite(path_out + filename, image)
tools_IO.save_mat(descript_ion, path_out + 'descript.ion', delim=' ')
return