计算机视觉-目标检测.zip

import os import cv2 import numpy as np import matplotlib.pyplot as plt # 加载预训练的人脸检测模型 face_cascade = cv2.CascadeClassifier('haarcascade_frontalface_default.xml') # 模板图像路径 cup_front_template_path = 'data/templates/cup-front.png' cup_up_template_path = 'data/templates/cup-up.png' cup_move_template_path = 'data/templates/cup-move.png' cup_down_template_path = 'data/templates/cup-down.png' cup_incline_template_path = 'data/templates/cup-incline.png' cup_incline_move_template_path = 'data/templates/cup-incline-move.png' cup_incline_up_template_path = 'data/templates/cup-incline-up.png' cup_incline_down_template_path = 'data/templates/cup-incline-down.png' cup_move_quick_template_path = 'data/templates/cup-move-quick.png' razor_front_template_path = 'data/templates/razor-front.png' razor_incline_template_path = 'data/templates/razor-up.png' razor_up_template_path = 'data/templates/razor-incline.png' razor_incline_up_template_path = 'data/templates/razor-incline-up.png' # 加载模板图像 cup_front_template = cv2.imread(cup_front_template_path, cv2.IMREAD_GRAYSCALE) cup_up_template = cv2.imread(cup_up_template_path, cv2.IMREAD_GRAYSCALE) cup_move_template = cv2.imread(cup_move_template_path, cv2.IMREAD_GRAYSCALE) cup_down_template = cv2.imread(cup_down_template_path, cv2.IMREAD_GRAYSCALE) cup_incline_template = cv2.imread(cup_incline_template_path, cv2.IMREAD_GRAYSCALE) cup_incline_move_template = cv2.imread(cup_incline_move_template_path, cv2.IMREAD_GRAYSCALE) cup_incline_up_template = cv2.imread(cup_incline_up_template_path, cv2.IMREAD_GRAYSCALE) cup_incline_down_template = cv2.imread(cup_incline_down_template_path, cv2.IMREAD_GRAYSCALE) cup_move_quick_template = cv2.imread(cup_move_quick_template_path, cv2.IMREAD_GRAYSCALE) razor_front_template = cv2.imread(razor_front_template_path, cv2.IMREAD_GRAYSCALE) razor_incline_template = cv2.imread(razor_incline_template_path, cv2.IMREAD_GRAYSCALE) razor_up_template = cv2.imread(razor_up_template_path, cv2.IMREAD_GRAYSCALE) razor_incline_up_template = cv2.imread(razor_incline_up_template_path, cv2.IMREAD_GRAYSCALE) # 定义目标检测函数（Haar级联分类器） def detect_faces(image, face_cascade, scale_factor=1.1, min_neighbors=5): gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) # 转换为灰度图像 faces = face_cascade.detectMultiScale(gray, scaleFactor=scale_factor, minNeighbors=min_neighbors) return faces # 非极大值抑制函数 def non_max_suppression(boxes, overlap_thresh=0.3): if len(boxes) == 0: return [] if boxes.dtype.kind == "i": boxes = boxes.astype("float") pick = [] x1 = boxes[:, 0] y1 = boxes[:, 1] x2 = boxes[:, 2] y2 = boxes[:, 3] scores = boxes[:, 4] area = (x2 - x1 + 1) * (y2 - y1 + 1) idxs = np.argsort(scores)[::-1] while len(idxs) > 0: last = idxs[0] pick.append(last) xx1 = np.maximum(x1[last], x1[idxs[1:]]) yy1 = np.maximum(y1[last], y1[idxs[1:]]) xx2 = np.minimum(x2[last], x2[idxs[1:]]) yy2 = np.minimum(y2[last], y2[idxs[1:]]) w = np.maximum(0, xx2 - xx1 + 1) h = np.maximum(0, yy2 - yy1 + 1) overlap = (w * h) / area[idxs[1:]] idxs = np.delete(idxs, np.concatenate(([0], np.where(overlap > overlap_thresh)[0] + 1))) return boxes[pick].astype("int") # 定义多尺度模板匹配函数 def multi_scale_template_matching(image, template, threshold=0.7, scale_range=(0.7, 1.3), scale_steps=6): gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) # 转换为灰度图像 h, w = template.shape[:2] matches = [] for scale in np.linspace(scale_range[0], scale_range[1], scale_steps): resized_template = cv2.resize(template, (int(w * scale), int(h * scale))) if resized_template.shape[0] > gray.shape[0] or resized_template.shape[1] > gray.shape[1]: continue res = cv2.matchTemplate(gray, resized_template, cv2.TM_CCOEFF_NORMED) loc = np.where(res >= threshold) for pt in zip(*loc[::-1]): matches.append( (pt[0], pt[1], pt[0] + resized_template.shape[1], pt[1] + resized_template.shape[0], res[pt[1], pt[0]])) # 应用非极大值抑制来过滤匹配结果 matches = np.array(matches) if len(matches) > 0: matches = non_max_suppression(matches, overlap_thresh=0.3) return matches # 待检测图像的目录 test_image_dir = 'data/q1-frame' output_dir = 'data/outframe' # 创建输出目录（如果不存在） os.makedirs(output_dir, exist_ok=True) def detect(test_image): # 检测人脸 faces = detect_faces(test_image, face_cascade) # 多尺度模板匹配水杯 threshold_second = 0.8 cup_front_matches = multi_scale_template_matching(test_image, cup_front_template) cup_up_matches = multi_scale_template_matching(test_image, cup_up_template) cup_move_matches = multi_scale_template_matching(test_image, cup_move_template) cup_down_matches = multi_scale_template_matching(test_image, cup_down_template, threshold_second) cup_incline_matches = multi_scale_template_matching(test_image, cup_incline_template, threshold_second) cup_incline_move_matches = multi_scale_template_matching(test_image, cup_incline_move_template, threshold_second) cup_incline_up_matches = multi_scale_template_matching(test_image, cup_incline_up_template, threshold_second) cup_incline_down_matches = multi_scale_template_matching(test_image, cup_incline_down_template, threshold_second) cup_move_quick_matches = multi_scale_template_matching(test_image, cup_move_quick_template, threshold_second) # 多尺度模板匹配书本 razor_front_matches = multi_scale_template_matching(test_image, razor_front_template) razor_incline_matches = multi_scale_template_matching(test_image, razor_incline_template) razor_up_matches = multi_scale_template_matching(test_image, razor_up_template) razor_incline_up_matches = multi_scale_template_matching(test_image, razor_incline_up_template) # # 绘制人脸边界框 # for (x, y, w, h) in faces: # cv2.rectangle(test_image, (x, y), (x + w, y + h), (0, 255, 0), 2) # cv2.putText(test_image, 'face', (x, y), cv2.FONT_HERSHEY_COMPLEX, 2, (0, 255, 0), 2) # # # 绘制水杯边界框 # for match in [cup_front_matches, cup_up_matches, cup_move_matches, cup_down_matches, cup_incline_matches, # cup_incline_move_matches, cup_incline_up_matches, cup_incline_down_matches, cup_move_quick_matches]: # if match.size > 0: # for (x1, y1, x2, y2, _) in match: # cv2.rectangle(test_image, (x1, y1), (x2, y2), (255, 0, 0), 2) # cv2.putText(test_image, 'cup', (x1, y1), cv2.FONT_HERSHEY_COMPLEX, 2, (0, 255, 0), 2) # # # 绘制刮胡刀边界框 # for match in [razor_front_matches, razor_up_matches, razor_incline_matches, razor_incline_up_matches]: # if match.size > 0: # for (x1, y1, x2, y2, _) in match: # cv2.rectangle(test_image, (x1, y1), (x2, y2), (0, 0, 255), 2) # cv2.putText(test_image, 'razor', (x1, y1), cv2.FONT_HERSHEY_COMPLEX, 2, (0, 255, 0), 2) # # 使用Matplotlib显示结果（可选） # plt.imshow(cv2.cvtColor(test_image, cv2.COLOR_BGR2RGB)) # plt.title(f'Detected Objects in {test_image}') # plt.axis('off') # plt.show() # 整合所有检测结果 # all_matches = [] # all_matches.extend([(x, y, w, h) for (x, y, w, h) in faces]) # 人脸检测结果 # for match_set in [cup_front_matches, cup_up_matches, cup_move_matches, cup_down_matches, cup_incline_matches, # cup_incline_move_matches, cup_incline_up_matches, cup_incline_down_matches, cup_move_quick_matches, # razor_front_matches, razor_up_matches, razor_incline_matches, razor_incline_up_matches]: # all_matches.extend([(x1, y