Vision AI 기반 제조 자동화 검수 시스템

# 주요 기술 구성
Object Detection: YOLOv5s      # 경량화된 실시간 용기 탐지
Object Tracking : DeepSORT     # 개별 용기 추적 및 ID 관리
Image Processing: OpenCV       # 영상 처리 및 전처리
GUI Framework   : PyQt5        # 독립 실행 데스크톱 앱
Deployment      : Pyinstaller  # .exe 패키징으로 로컬 대응
Data Processing : NumPy, Pandas

# 데이터 수집 전략
data_collection_strategy = {
    "영상 수": 50개,
    "프레임 추출": "약 400장 BBox 라벨링",
    "환경 조건": [
        "아침/점심/저녁 조명 변화",
        "재료 적재 상태 변화", 
        "제조 카운터 위치별 각도",
        "용기 종류별 (빈 용기 ~ 완성품)"
    ]
}

# YOLOv5s 학습 설정
training_config = {
    "model": "YOLOv5s",           # 실시간 처리를 위한 경량 모델
    "epochs": 300,
    "batch_size": 16,
    "img_size": 640,
    "data": "container_dataset.yaml",
    "weights": "yolov5s.pt"       # 사전 학습된 가중치
}

# 데이터셋 구성
dataset_structure = {
    "train": "400장 (80%)",
    "val": "100장 (20%)",
    "classes": ["empty_container", "filled_container", "completed_product"]
}

# OCP, 객체 중앙점 설정
    def box_label(self, box, label='', color=(128, 128, 128), txt_color=(255, 255, 255)):
        # Add one xyxy box to image with label
        if self.pil or not is_ascii(label):
            self.draw.rectangle(box, width=self.lw, outline=color)  # box
            if label:
                w, h = self.font.getsize(label)  # text width, height
                outside = box[1] - h >= 0  # label fits outside box
                self.draw.rectangle(
                    (box[0], box[1] - h if outside else box[1], box[0] + w + 1,
                     box[1] + 1 if outside else box[1] + h + 1),
                    fill=color,
                )
                # self.draw.text((box[0], box[1]), label, fill=txt_color, font=self.font, anchor='ls')  # for PIL>8.0
                self.draw.text((box[0], box[1] - h if outside else box[1]), label, fill=txt_color, font=self.font)
        else:  # cv2
            p1, p2 = (int(box[0]), int(box[1])), (int(box[2]), int(box[3]))
            center_coordinates = (int(box[0] + (box[2]-box[0])/2), int(box[1] + (box[3] - box[1])/2))
            cv2.rectangle(self.im, p1, p2, color, thickness=self.lw, lineType=cv2.LINE_AA)
            cv2.circle(self.im, center_coordinates, radius=3, color=color, thickness=3)
            if label:
                tf = max(self.lw - 1, 1)  # font thickness
                w, h = cv2.getTextSize(label, 0, fontScale=self.lw / 3, thickness=tf)[0]  # text width, height
                outside = p1[1] - h >= 3
                p2 = p1[0] + w, p1[1] - h - 3 if outside else p1[1] + h + 3
                cv2.rectangle(self.im, p1, p2, color, -1, cv2.LINE_AA)  # filled
                cv2.putText(self.im,
                            label, (p1[0], p1[1] - 2 if outside else p1[1] + h + 2),
                            0,
                            self.lw / 3,
                            txt_color,
                            thickness=tf,
                            lineType=cv2.LINE_AA

#개수판단
class Count:
    def count_1_function(det, im, s, im0, names, outputs, tracker_list, dt, i, t3,t2,tracking_method,annotator, save_txt, txt_path,frame_idx, save_vid, save_crop, show_vid, hide_labels, hide_conf, hide_class, path, imc, save_dir, p):
        w, h = im0.shape[1], im0.shape[0]
        if det is not None and len(det):
            det[:, :4] = scale_coords(im.shape[2:], det[:, :4], im0.shape).round()
            for c in det[:, -1].unique():
                n = (det[:, -1] == c).sum()  # detections per class
                s += f"{n} {names[int(c)]}{'s' * (n > 1)}, "
            t4 = time_sync()
            outputs[i] = tracker_list[i].update(det.cpu(), im0)
            t5 = time_sync()
            dt[3] += t5 - t4
            if len(outputs[i]) > 0:
                for j, (output, conf) in enumerate(zip(outputs[i], det[:, 4])):
                    bboxes = output[0:4]
                    id = output[4]
                    cls = output[5]
                    global count_web_1, data_web_1, order_index_web_1, step_count_web_1
                    center_coordinates = (
                        int(bboxes[0] + (bboxes[2]-bboxes[0])/2), int(bboxes[1] + (bboxes[3] - bboxes[1])/2))
                    
                    if (int(bboxes[0]+(bboxes[2] - bboxes[0])/2) < (int(w/2))) and (id not in data_web_1):

                        # like sensor
                        im0 = cv2.rectangle(im0, (0,0), (w,h), (0,0,255), -1)
                        
                        count_web_1 += 1
                        data_web_1.append(id)
                        order_data_count = order_data['Count']
                        step_count_web_1 += 1
                        
                        if step_count_web_1 >= int(order_data_count[order_index_web_1]):
                            order_index_web_1+= 1
                            step_count_web_1 = 0 
                            
                    common_save_functions(output, save_txt, txt_path, frame_idx, i, save_vid, save_crop, show_vid, id, cls,hide_labels, names, hide_conf, conf, hide_class, annotator, bboxes, path, imc, save_dir, p)
                    
            # LOGGER.info(f'{s}Done. yolo:({t3 - t2:.3f}s), {tracking_method}:({t5 - t4:.3f}s)')
        else:
            # strongsort_list[i].increment_ages()
            # LOGGER.info('No detections')    
            pass
        im0 = annotator.result()
        return im0, count_web_1, order_index_web_1, step_count_web_1