Spaces:
Runtime error
Runtime error
File size: 7,747 Bytes
c75618b 6bd3d08 c75618b 6bd3d08 c75618b 6bd3d08 c75618b 971df0f 6bd3d08 971df0f 6bd3d08 971df0f 6bd3d08 971df0f 6bd3d08 971df0f 6bd3d08 971df0f 6bd3d08 971df0f 6bd3d08 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 | # -----------------------------------------------------------------------------
#
# This file is part of the PantoScanner distribution on:
# https://huggingface.co/spaces/swissrail/PantoScanner
#
# PantoScanner - Analytics and measurement capability for technical objects.
# Copyright (C) 2017-2024 Schweizerische Bundesbahnen SBB
#
# Authors (C) 2024 L. Hofstetter (lukas.hofstetter@sbb.ch)
# Authors (C) 2017 U. Gehrig (urs.gehrig@sbb.ch)
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program. If not, see <https://www.gnu.org/licenses/>.
#
# -----------------------------------------------------------------------------
import os
from glob import glob # FIX: glob2 is a Python 2 backport; use stdlib glob
import torch
import torchvision
import cv2 as cv
import matplotlib.pyplot as plt
import matplotlib
import matplotlib.patches as patches
import numpy as np
from dataclasses import dataclass, field, asdict, astuple, InitVar
from typing import List, Dict, Tuple
@dataclass
class BBoxCoordinates:
x_min: int = field(init=True, default=None)
x_max: int = field(init=True, default=None)
y_min: int = field(init=True, default=None)
y_max: int = field(init=True, default=None)
x_center: float = field(init=False, default=False)
y_center: float = field(init=False, default=False)
height: int = field(init=False, default=False)
width: int = field(init=False, default=False)
as_array: np.ndarray = field(init=False, default=False)
def __post_init__(self):
self.height = self.x_max - self.x_min
self.width = self.y_max - self.y_min
self.x_center = 0.5 * (self.x_max + self.x_min)
self.y_center = 0.5 * (self.y_max + self.y_min)
def shift_coordinates(self, x_shift, y_shift, x_new_low_bound=None, x_new_upp_bound=None,
y_new_low_bound=None, y_new_upp_bound=None):
self.x_min = self.x_min - x_shift
if x_new_low_bound is not None:
self.x_min = max(self.x_min, x_new_low_bound)
self.x_max = self.x_max - x_shift
if x_new_upp_bound is not None:
self.x_max = min(self.x_max, x_new_upp_bound)
self.y_min = self.y_min - y_shift
if y_new_low_bound is not None:
self.y_min = max(self.y_min, y_new_low_bound)
self.y_max = self.y_max - y_shift
if y_new_upp_bound is not None:
self.y_max = min(self.y_max, y_new_upp_bound)
self.height = self.x_max - self.x_min
self.width = self.y_max - self.y_min
self.x_center = 0.5 * (self.x_max + self.x_min)
self.y_center = 0.5 * (self.y_max + self.y_min)
def check_valid(self):
if self.width and self.height:
return True
else:
return False
def make_yolo_label_string(self, label_number, img_size, float_precision):
final_string = str(int(label_number))
for this_float in [self.y_center, self.x_center, self.width, self.height]:
final_string = final_string + ' ' + '{:.{n}f}'.format(this_float/img_size, n=float_precision)
return final_string
def evaluate_yolo(model_path: str, img_path: str, img_size: int, bound_1: Tuple[int, int], bound_2: Tuple[int, int]):
model = torch.hub.load('ultralytics/yolov5', 'custom', path=model_path, verbose=None) # local model
inpt_img = cv.imread(img_path)
img_1 = inpt_img[bound_1[0]:bound_1[0] + img_size, bound_1[1]:bound_1[1] + img_size, :]
img_2 = inpt_img[bound_2[0]:bound_2[0] + img_size, bound_2[1]:bound_2[1] + img_size, :]
inpt_imgs = [img_1, img_2]
results = model(inpt_imgs, size=img_size)
result_list_images = []
results_1 = results.pandas().xyxy[0].reset_index() # img1 predictions (pandas)
results_2 = results.pandas().xyxy[1].reset_index() # img1 predictions (pandas)
for index, row in results_1.iterrows():
result_list_images.append(
[row['xmin'] + bound_1[1], row['xmax'] + bound_1[1], row['ymin'] + bound_1[0], row['ymax'] + bound_1[0]])
for index, row in results_2.iterrows():
result_list_images.append(
[row['xmin'] + bound_2[1], row['xmax'] + bound_2[1], row['ymin'] + bound_2[0], row['ymax'] + bound_2[0]])
result_arr = np.asarray(result_list_images)
x_min = np.min(result_arr[:, 0])
x_max = np.max(result_arr[:, 1])
y_min = np.min(result_arr[:, 2])
y_max = np.max(result_arr[:, 3])
bound_final_x = max(min(int(0.5 * (x_min + x_max - img_size)), 3208 - img_size), 0)
bound_final_y = max(min(int(0.5 * (y_min + y_max - img_size)), 2200 - img_size), 0)
img_final = inpt_img[bound_final_y:bound_final_y + img_size, bound_final_x:bound_final_x + img_size, :]
imgs_final = [img_final]
results_final = model(imgs_final, size=img_size)
bboxes_final = results_final.pandas().xyxy[0].reset_index() # img1 predictions (pandas)
final_bbox_list = []
for index, row in bboxes_final.iterrows():
this_box = BBoxCoordinates(row['ymin'] + bound_final_y, row['ymax'] + bound_final_y,
row['xmin'] + bound_final_x, row['xmax'] + bound_final_x)
final_bbox_list.append(this_box)
return final_bbox_list
def evaluate_yolo_2(model_loaded, img_path: str, img_size: int, bound_1: Tuple[int, int], bound_2: Tuple[int, int]):
model = model_loaded
inpt_img = cv.imread(img_path)
img_1 = inpt_img[bound_1[0]:bound_1[0] + img_size, bound_1[1]:bound_1[1] + img_size, :]
img_2 = inpt_img[bound_2[0]:bound_2[0] + img_size, bound_2[1]:bound_2[1] + img_size, :]
inpt_imgs = [img_1, img_2]
results = model(inpt_imgs, size=img_size)
result_list_images = []
results_1 = results.pandas().xyxy[0].reset_index() # img1 predictions (pandas)
results_2 = results.pandas().xyxy[1].reset_index() # img1 predictions (pandas)
for index, row in results_1.iterrows():
result_list_images.append(
[row['xmin'] + bound_1[1], row['xmax'] + bound_1[1], row['ymin'] + bound_1[0], row['ymax'] + bound_1[0]])
for index, row in results_2.iterrows():
result_list_images.append(
[row['xmin'] + bound_2[1], row['xmax'] + bound_2[1], row['ymin'] + bound_2[0], row['ymax'] + bound_2[0]])
result_arr = np.asarray(result_list_images)
x_min = np.min(result_arr[:, 0])
x_max = np.max(result_arr[:, 1])
y_min = np.min(result_arr[:, 2])
y_max = np.max(result_arr[:, 3])
bound_final_x = max(min(int(0.5 * (x_min + x_max - img_size)), 3208 - img_size), 0)
bound_final_y = max(min(int(0.5 * (y_min + y_max - img_size)), 2200 - img_size), 0)
img_final = inpt_img[bound_final_y:bound_final_y + img_size, bound_final_x:bound_final_x + img_size, :]
imgs_final = [img_final]
results_final = model(imgs_final, size=img_size)
bboxes_final = results_final.pandas().xyxy[0].reset_index() # img1 predictions (pandas)
final_bbox_list = []
for index, row in bboxes_final.iterrows():
this_box = BBoxCoordinates(row['ymin'] + bound_final_y, row['ymax'] + bound_final_y,
row['xmin'] + bound_final_x, row['xmax'] + bound_final_x)
final_bbox_list.append(this_box)
return final_bbox_list |