Spaces:
Sleeping
Sleeping
File size: 3,429 Bytes
5d92054 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 |
# File: processing.pyx
# cython: language_level=3, boundscheck=False, wraparound=False, nonecheck=False, cdivision=True
import numpy as np
cimport numpy as np
import pandas as pd
cimport pandas as pd
from libc.math cimport isnan
from cpython.mem cimport PyMem_Malloc, PyMem_Free
# Define C types for better performance
ctypedef np.float32_t DTYPE_t
ctypedef np.int32_t ITYPE_t
def process_bbox_depth_cy(np.ndarray[DTYPE_t, ndim=2] depth_map,
int y_min, int y_max, int x_min, int x_max):
"""
Optimized bbox depth calculations using Cython
"""
cdef:
int i, j, count = 0
double sum_val = 0.0
double mean_val = 0.0
np.ndarray[DTYPE_t, ndim=1] flat_vals
int flat_size = 0
for i in range(y_min, y_max):
for j in range(x_min, x_max):
if not isnan(depth_map[i, j]):
sum_val += depth_map[i, j]
count += 1
if count > 0:
mean_val = sum_val / count
# Create array for trimmed mean calculation
flat_vals = np.zeros(count, dtype=np.float32)
flat_size = 0
for i in range(y_min, y_max):
for j in range(x_min, x_max):
if not isnan(depth_map[i, j]):
flat_vals[flat_size] = depth_map[i, j]
flat_size += 1
return mean_val, np.median(flat_vals), np.percentile(flat_vals, [20, 80])
def handle_overlaps_cy(np.ndarray[DTYPE_t, ndim=2] depth_map,
np.ndarray[ITYPE_t, ndim=2] boxes):
"""
Optimized overlap handling using Cython
"""
cdef:
int n_boxes = boxes.shape[0]
int i, j
int y_min1, y_max1, x_min1, x_max1
int y_min2, y_max2, x_min2, x_max2
double area1, area2, area_intersection
bint* to_remove = <bint*>PyMem_Malloc(n_boxes * sizeof(bint))
if not to_remove:
raise MemoryError()
try:
for i in range(n_boxes):
to_remove[i] = False
for i in range(n_boxes):
if to_remove[i]:
continue
y_min1, x_min1, y_max1, x_max1 = boxes[i]
for j in range(i + 1, n_boxes):
if to_remove[j]:
continue
y_min2, x_min2, y_max2, x_max2 = boxes[j]
# Calculate intersection
y_min_int = max(y_min1, y_min2)
y_max_int = min(y_max1, y_max2)
x_min_int = max(x_min1, x_min2)
x_max_int = min(x_max1, x_max2)
if y_min_int < y_max_int and x_min_int < x_max_int:
area1 = (y_max1 - y_min1) * (x_max1 - x_min1)
area2 = (y_max2 - y_min2) * (x_max2 - x_min2)
area_intersection = (y_max_int - y_min_int) * (x_max_int - x_min_int)
if area_intersection / min(area1, area2) >= 0.70:
if area1 < area2:
to_remove[i] = True
break
else:
to_remove[j] = True
return np.array([i for i in range(n_boxes) if not to_remove[i]], dtype=np.int32)
finally:
PyMem_Free(to_remove)
|