5149989027

pending completion

Build # 5149989027

Build Type

push

github

Committed by

web-flow

Commit Message

Merge branch 'main' into staging

Run Details

684 of 1187 relevant lines covered (57.62%)

0.58 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

69.12

/src/dfp/utils/util.py

import cv2
import numpy as np
import tensorflow as tf
from scipy import ndimage


def fast_hist(im: np.ndarray, gt: np.ndarray, n: int = 9) -> np.ndarray:
    """
    n is num_of_classes
    """
    k = (gt >= 0) & (gt < n)
    return np.bincount(
        n * gt[k].astype(int) + im[k], minlength=n**2
    ).reshape(n, n)


def flood_fill(test_array: np.ndarray, h_max: int = 255) -> np.ndarray:
    """
    fill in the hole
    """
    test_array = test_array.squeeze()
    input_array = np.copy(test_array)
    el = ndimage.generate_binary_structure(2, 2).astype(int)
    inside_mask = ndimage.binary_erosion(~np.isnan(input_array), structure=el)
    output_array = np.copy(input_array)
    output_array[inside_mask] = h_max
    output_old_array = np.copy(input_array)
    output_old_array.fill(0)
    el = ndimage.generate_binary_structure(2, 1).astype(int)
    while not np.array_equal(output_old_array, output_array):
        output_old_array = np.copy(output_array)
        output_array = np.maximum(
            input_array,
            ndimage.grey_erosion(output_array, size=(3, 3), footprint=el),
        )
    return output_array


def fill_break_line(cw_mask: np.ndarray) -> np.ndarray:
    broken_line_h = np.array(
        [
            [0, 0, 0, 0, 0],
            [0, 0, 0, 0, 0],
            [1, 0, 0, 0, 1],
            [0, 0, 0, 0, 0],
            [0, 0, 0, 0, 0],
        ],
        dtype=np.uint8,
    )
    broken_line_h2 = np.array(
        [
            [0, 0, 0, 0, 0],
            [0, 0, 0, 0, 0],
            [1, 1, 0, 1, 1],
            [0, 0, 0, 0, 0],
            [0, 0, 0, 0, 0],
        ],
        dtype=np.uint8,
    )
    broken_line_v = np.transpose(broken_line_h)
    broken_line_v2 = np.transpose(broken_line_h2)
    cw_mask = cv2.morphologyEx(cw_mask, cv2.MORPH_CLOSE, broken_line_h)
    cw_mask = cv2.morphologyEx(cw_mask, cv2.MORPH_CLOSE, broken_line_v)
    cw_mask = cv2.morphologyEx(cw_mask, cv2.MORPH_CLOSE, broken_line_h2)
    cw_mask = cv2.morphologyEx(cw_mask, cv2.MORPH_CLOSE, broken_line_v2)

    return cw_mask


def refine_room_region(cw_mask: np.ndarray, rm_ind: np.ndarray) -> np.ndarray:
    label_rm, num_label = ndimage.label((1 - cw_mask))
    new_rm_ind = np.zeros(rm_ind.shape)
    for j in range(1, num_label + 1):
        mask = (label_rm == j).astype(np.uint8)
        ys, xs, _ = np.where(mask != 0)
        area = (np.amax(xs) - np.amin(xs)) * (np.amax(ys) - np.amin(ys))
        if area < 100:
            continue
        else:
            room_types, type_counts = np.unique(
                mask * rm_ind, return_counts=True
            )
            if len(room_types) > 1:
                room_types = room_types[1:]
                # ignore background type which is zero
                type_counts = type_counts[1:]
                # ignore background count
            new_rm_ind += mask * room_types[np.argmax(type_counts)]

    return new_rm_ind


def print_model_weights_sparsity(model):
    for layer in model.layers:
        if isinstance(layer, tf.keras.layers.Wrapper):
            weights = layer.trainable_weights
        else:
            weights = layer.weights
        for weight in weights:
            if "kernel" not in weight.name or "centroid" in weight.name:
                continue
            weight_size = weight.numpy().size
            zero_num = np.count_nonzero(weight == 0)
            print(
                f"{weight.name}: {zero_num/weight_size:.2%} sparsity ",
                f"({zero_num}/{weight_size})",
            )


def print_model_weight_clusters(model):
    for layer in model.layers:
        if isinstance(layer, tf.keras.layers.Wrapper):
            weights = layer.trainable_weights
        else:
            weights = layer.weights
        for weight in weights:
            # ignore auxiliary quantization weights
            if "quantize_layer" in weight.name:
                continue
            if "kernel" in weight.name:
                unique_count = len(np.unique(weight))
                print(f"{layer.name}/{weight.name}: {unique_count} clusters ")

1	import cv2	1✔
2	import numpy as np	1✔
3	import tensorflow as tf	1✔
4	from scipy import ndimage	1✔
5
6
7	def fast_hist(im: np.ndarray, gt: np.ndarray, n: int = 9) -> np.ndarray:	1✔
8	"""
9	n is num_of_classes
10	"""
11	k = (gt >= 0) & (gt < n)	1✔
12	return np.bincount(	1✔
13	n * gt[k].astype(int) + im[k], minlength=n**2
14	).reshape(n, n)
15
16
17	def flood_fill(test_array: np.ndarray, h_max: int = 255) -> np.ndarray:	1✔
18	"""
19	fill in the hole
20	"""
21	test_array = test_array.squeeze()	1✔
22	input_array = np.copy(test_array)	1✔
23	el = ndimage.generate_binary_structure(2, 2).astype(int)	1✔
24	inside_mask = ndimage.binary_erosion(~np.isnan(input_array), structure=el)	1✔
25	output_array = np.copy(input_array)	1✔
26	output_array[inside_mask] = h_max	1✔
27	output_old_array = np.copy(input_array)	1✔
28	output_old_array.fill(0)	1✔
29	el = ndimage.generate_binary_structure(2, 1).astype(int)	1✔
30	while not np.array_equal(output_old_array, output_array):	1✔
31	output_old_array = np.copy(output_array)	1✔
32	output_array = np.maximum(	1✔
33	input_array,
34	ndimage.grey_erosion(output_array, size=(3, 3), footprint=el),
35	)
36	return output_array	1✔
37
38
39	def fill_break_line(cw_mask: np.ndarray) -> np.ndarray:	1✔
40	broken_line_h = np.array(	1✔
41	[
42	[0, 0, 0, 0, 0],
43	[0, 0, 0, 0, 0],
44	[1, 0, 0, 0, 1],
45	[0, 0, 0, 0, 0],
46	[0, 0, 0, 0, 0],
47	],
48	dtype=np.uint8,
49	)
50	broken_line_h2 = np.array(	1✔
51	[
52	[0, 0, 0, 0, 0],
53	[0, 0, 0, 0, 0],
54	[1, 1, 0, 1, 1],
55	[0, 0, 0, 0, 0],
56	[0, 0, 0, 0, 0],
57	],
58	dtype=np.uint8,
59	)
60	broken_line_v = np.transpose(broken_line_h)	1✔
61	broken_line_v2 = np.transpose(broken_line_h2)	1✔
62	cw_mask = cv2.morphologyEx(cw_mask, cv2.MORPH_CLOSE, broken_line_h)	1✔
63	cw_mask = cv2.morphologyEx(cw_mask, cv2.MORPH_CLOSE, broken_line_v)	1✔
64	cw_mask = cv2.morphologyEx(cw_mask, cv2.MORPH_CLOSE, broken_line_h2)	1✔
65	cw_mask = cv2.morphologyEx(cw_mask, cv2.MORPH_CLOSE, broken_line_v2)	1✔
66
67	return cw_mask	1✔
68
69
70	def refine_room_region(cw_mask: np.ndarray, rm_ind: np.ndarray) -> np.ndarray:	1✔
71	label_rm, num_label = ndimage.label((1 - cw_mask))	1✔
72	new_rm_ind = np.zeros(rm_ind.shape)	1✔
73	for j in range(1, num_label + 1):	1✔
74	mask = (label_rm == j).astype(np.uint8)	1✔
75	ys, xs, _ = np.where(mask != 0)	1✔
76	area = (np.amax(xs) - np.amin(xs)) * (np.amax(ys) - np.amin(ys))	1✔
77	if area < 100:	1✔
78	continue	×
79	else:
80	room_types, type_counts = np.unique(	1✔
81	mask * rm_ind, return_counts=True
82	)
83	if len(room_types) > 1:	1✔
84	room_types = room_types[1:]	1✔
85	# ignore background type which is zero
86	type_counts = type_counts[1:]	1✔
87	# ignore background count
88	new_rm_ind += mask * room_types[np.argmax(type_counts)]	1✔
89
90	return new_rm_ind	1✔
91
92
93	def print_model_weights_sparsity(model):	1✔
94	for layer in model.layers:	×
95	if isinstance(layer, tf.keras.layers.Wrapper):	×
96	weights = layer.trainable_weights	×
97	else:
98	weights = layer.weights	×
99	for weight in weights:	×
100	if "kernel" not in weight.name or "centroid" in weight.name:	×
101	continue	×
102	weight_size = weight.numpy().size	×
103	zero_num = np.count_nonzero(weight == 0)	×
104	print(	×
105	f"{weight.name}: {zero_num/weight_size:.2%} sparsity ",
106	f"({zero_num}/{weight_size})",
107	)
108
109
110	def print_model_weight_clusters(model):	1✔
111	for layer in model.layers:	×
112	if isinstance(layer, tf.keras.layers.Wrapper):	×
113	weights = layer.trainable_weights	×
114	else:
115	weights = layer.weights	×
116	for weight in weights:	×
117	# ignore auxiliary quantization weights
118	if "quantize_layer" in weight.name:	×
119	continue	×
120	if "kernel" in weight.name:	×
121	unique_count = len(np.unique(weight))	×
122	print(f"{layer.name}/{weight.name}: {unique_count} clusters ")	×

zcemycl / TF2DeepFloorplan / 5149989027

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous