안녕하세요. 별도의 서버에서 pretrained 모델 기반으로 video inference 실행을 하는 중에서 계속 다음과 같은 에러가 발생하여 문의드립니다. colab이 아닌 별도 서버에서 실행하였고, 선생님이 주신 코드를 파일 경로 부분만 수정해서 사용하였습니다. ------------------- import sys sys.path.append("/home/jongmin/mmdetection") # config 파일과 pretrained 모델을 기반으로 Detector 모델을 생성.  from mmdet.apis import init_detector, inference_detector import cv2 import matplotlib.pyplot as plt labels_to_names_seq = {0:'person',1:'bicycle',2:'car',3:'motorbike',4:'aeroplane',5:'bus',6:'train',7:'truck',8:'boat',9:'traffic light',10:'fire hydrant',                         11:'stop sign',12:'parking meter',13:'bench',14:'bird',15:'cat',16:'dog',17:'horse',18:'sheep',19:'cow',20:'elephant',                         21:'bear',22:'zebra',23:'giraffe',24:'backpack',25:'umbrella',26:'handbag',27:'tie',28:'suitcase',29:'frisbee',30:'skis',                         31:'snowboard',32:'sports ball',33:'kite',34:'baseball bat',35:'baseball glove',36:'skateboard',37:'surfboard',38:'tennis racket',39:'bottle',40:'wine glass',                         41:'cup',42:'fork',43:'knife',44:'spoon',45:'bowl',46:'banana',47:'apple',48:'sandwich',49:'orange',50:'broccoli',                         51:'carrot',52:'hot dog',53:'pizza',54:'donut',55:'cake',56:'chair',57:'sofa',58:'pottedplant',59:'bed',60:'diningtable',                         61:'toilet',62:'tvmonitor',63:'laptop',64:'mouse',65:'remote',66:'keyboard',67:'cell phone',68:'microwave',69:'oven',70:'toaster',                         71:'sink',72:'refrigerator',73:'book',74:'clock',75:'vase',76:'scissors',77:'teddy bear',78:'hair drier',79:'toothbrush' } def get_detected_img(model, img_array,  score_threshold=0.3, is_print=True):       # 인자로 들어온 image_array를 복사.    draw_img = img_array.copy()   bbox_color=(0, 255, 0)   text_color=(0, 0, 255)   # model과 image array를 입력 인자로 inference detection 수행하고 결과를 results로 받음.    # results는 80개의 2차원 array(shape=(오브젝트갯수, 5))를 가지는 list.    results = inference_detector(model, img_array)   # 80개의 array원소를 가지는 results 리스트를 loop를 돌면서 개별 2차원 array들을 추출하고 이를 기반으로 이미지 시각화    # results 리스트의 위치 index가 바로 COCO 매핑된 Class id. 여기서는 result_ind가 class id   # 개별 2차원 array에 오브젝트별 좌표와 class confidence score 값을 가짐.    for result_ind, result in enumerate(results):     # 개별 2차원 array의 row size가 0 이면 해당 Class id로 값이 없으므로 다음 loop로 진행.      if len(result) == 0:       continue          # 2차원 array에서 5번째 컬럼에 해당하는 값이 score threshold이며 이 값이 함수 인자로 들어온 score_threshold 보다 낮은 경우는 제외.      result_filtered = result[np.where(result[:, 4] > score_threshold)]          # 해당 클래스 별로 Detect된 여러개의 오브젝트 정보가 2차원 array에 담겨 있으며, 이 2차원 array를 row수만큼 iteration해서 개별 오브젝트의 좌표값 추출.      for i in range(len(result_filtered)):       # 좌상단, 우하단 좌표 추출.        left = int(result_filtered[i, 0])       top = int(result_filtered[i, 1])       right = int(result_filtered[i, 2])       bottom = int(result_filtered[i, 3])       caption = "{}: {:.4f}".format(labels_to_names_seq[result_ind], result_filtered[i, 4])       cv2.rectangle(draw_img, (left, top), (right, bottom), color=bbox_color, thickness=2)       cv2.putText(draw_img, caption, (int(left), int(top - 7)), cv2.FONT_HERSHEY_SIMPLEX, 0.37, text_color, 1)       if is_print:         print(caption)   return draw_img # config 파일을 설정하고, 다운로드 받은 pretrained 모델을 checkpoint로 설정.  config_file = './configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py' checkpoint_file = './mmdetection/checkpoints/faster_rcnn_r50_fpn_1x_coco_20200130-047c8118.pth' model = init_detector(config_file, checkpoint_file, device='cuda:2') img = '/demo/demo.jpg' img_arr  = cv2.imread(img) detected_img = get_detected_img(model, img_arr,  score_threshold=0.5, is_print=True) # detect 입력된 이미지는 bgr임. 이를 최종 출력시 rgb로 변환  detected_img = cv2.cvtColor(detected_img, cv2.COLOR_BGR2RGB) plt.figure(figsize=(12, 12)) plt.imshow(detected_img) ------------------------------아래 내용입니다. 해당 내용에서 대해서 검색해보았을 때는 잘못된 메모리에 접근해서 그렇다고 하는데.. 잘 이해가 되지 않아 문의드립니다.  RuntimeError: CUDA error: invalid device function Segmentation fault (core dumped)