YOLOv7-obb上传kaggle遇到的问题合集_yolov7 读取kaggle数据集

一、utils.nms_rotated问题

cannot import name ‘nms_rotated_ext‘ from partially initialized module ‘utils.nms_rotated‘ 

 在kaggle上再重新进行一次编译即可

cd /kaggle/working/yolov7-obb/utils/nms_rotated
 
!python setup.py build_ext --inplace 

二、THC/THC.h问题

fatal error: THC/THC.h: No such file or directory

 运行问题一中的代码后随即出现该问题，因为pytorch1.10以后无THC.h头文件

参考代码

将poly_nms_cuda.cu替换为了链接中的代码，即可正常运行，如果pytorch版本在1.10以内则无需替换，但是显卡对pytorch版本有要求，需注意~

三、图片格式问题

原始数据集为.png格式，忘记进行转换了

参考代码

亲测好用~

四、训练中get_map问题

indices should be either on cpu or on the same device as the indexed tensor (cpu)

五、utils_map.py问题

NameError: name 'COCO' is not defined

ValueError: could not convert string to float: 'P-3C'

前者是未下载pycocotools，后者是utils_map.py中的obb的形式问题

       obb有两种形式：① (x,y,w,h,θ)       ②(x1,y1,x2,y2,x3,y3,x4,y4)

       utils_map.py和数据集的形式要对应，在utils_map.py中修改三个地方即可

#-------------- (x,y,w,h,θ) ----------------#
            # try:
            #     if "difficult" in line:
            #         class_name, x, y, w, h,angle, _difficult = line.split()
            #         is_difficult = True
            #     else:
            #         class_name, x, y, w, h,angle = line.split()
            # except:
            #     if "difficult" in line:
            #         line_split  = line.split()
            #         _difficult  = line_split[-1]
            #         angle       = line_split[-2]
            #         h           = line_split[-3]
            #         w           = line_split[-4]
            #         y           = line_split[-5]
            #         x           = line_split[-6]
            #         class_name  = ""
            #         for name in line_split[:-6]:
            #             class_name += name + " "
            #         class_name   = class_name[:-1]
            #         is_difficult = True
            #     else:
            #         line_split  = line.split()
            #         angle       = line_split[-1]
            #         h           = line_split[-2]
            #         w           = line_split[-3]
            #         y           = line_split[-4]
            #         x           = line_split[-5]
            #         class_name  = ""
            #         for name in line_split[:-5]:
            #             class_name += name + " "
            #         class_name = class_name[:-1]
            #
            # bbox = x + " " + y + " " + w + " " + h + " " + angle
 
 
#------------- (x1,y1,x2,y2,x3,y3,x4,y4) -----------------#
            try:
                if "difficult" in line:
                    class_name, left, top, right, bottom, _difficult = line.split()
                    is_difficult = True
                else:
                    class_name, left, top, right, bottom = line.split()
            except:
                if "difficult" in line:
                    line_split = line.split()
                    _difficult = line_split[-1]
                    bottom = line_split[-2]
                    right = line_split[-3]
                    top = line_split[-4]
                    left = line_split[-5]
                    class_name = ""
                    for name in line_split[:-5]:
                        class_name += name + " "
                    class_name = class_name[:-1]
                    is_difficult = True
                else:
                    line_split = line.split()
                    bottom = line_split[-1]
                    right = line_split[-2]
                    top = line_split[-3]
                    left = line_split[-4]
                    class_name = ""
                    for name in line_split[:-4]:
                        class_name += name + " "
                    class_name = class_name[:-1]
            bbox = left + " " + top + " " + right + " " + bottom

#--------------- (x,y,w,h,θ) ------------------#
# try:
#     tmp_class_name, confidence, x, y, w, h,angle = line.split()
# except:
#     line_split      = line.split()
#     angle           = line_split[-1]
#     h               = line_split[-2]
#     w               = line_split[-3]
#     y               = line_split[-4]
#     x               = line_split[-5]
#     confidence      = line_split[-6]
#     tmp_class_name  = ""
#     for name in line_split[:-6]:
#         tmp_class_name += name + " "
#     tmp_class_name  = tmp_class_name[:-1]
#
# if tmp_class_name == class_name:
#     bbox = x + " " + y + " " + w + " " + h + " " + angle
#     bounding_boxes.append({"confidence":confidence, "file_id":file_id, "bbox":bbox})
 
 
 
#--------------- (x1,y1,x2,y2,x3,y3,x4,y4) ------------------#
try:
    tmp_class_name, confidence, left, top, right, bottom = line.split()
except:
    line_split      = line.split()
    bottom          = line_split[-1]
    right           = line_split[-2]
    top             = line_split[-3]
    left            = line_split[-4]
    confidence      = line_split[-5]
    tmp_class_name  = ""
    for name in line_split[:-5]:
      tmp_class_name += name + " "
    tmp_class_name  = tmp_class_name[:-1]
 
if tmp_class_name == class_name:
    bbox = left + " " + top + " " + right + " " +bottom
    bounding_boxes.append({"confidence":confidence, "file_id":file_id, "bbox":bbox})

 for obj in ground_truth_data:
 
#--------------- (x,y,w,h,θ) ------------------#
  # if obj["class_name"] == class_name:
  #     bbgt    = [float(x) for x in obj["bbox"].split() ]
  #     box1    = np.array([bb[0], bb[1], bb[2], bb[3], bb[4]], np.float32)
  #     box2    = np.array([bbgt[0], bbgt[1], bbgt[2], bbgt[3], bbgt[4]], np.float32)
  #     ov      = iou_rotate_calculate(box1, box2)
  #     if ov > ovmax:
  #         ovmax = ov
  #         gt_match = obj
 
 
#--------------- (x1,y1,x2,y2,x3,y3,x4,y4) ------------------#
    if obj["class_name"] == class_name:
        bbgt    = [ float(x) for x in obj["bbox"].split() ]
        bi      = [max(bb[0],bbgt[0]), max(bb[1],bbgt[1]), min(bb[2],bbgt[2]), min(bb[3],bbgt[3])]
        iw      = bi[2] - bi[0] + 1
        ih      = bi[3] - bi[1] + 1
        if iw > 0 and ih > 0:
           ua = (bb[2] - bb[0] + 1) * (bb[3] - bb[1] + 1) + (bbgt[2] - bbgt[0]
                                            + 1) * (bbgt[3] - bbgt[1] + 1) - iw * ih
           ov = iw * ih / ua
           if ov > ovmax:
              ovmax = ov
              gt_match = obj

其中，iou_rotate_calculate对①是必需的，代码如下：

# def iou_rotate_calculate(boxes1, boxes2):
#     """
#     计算旋转面积
#     boxes1,boxes2格式为x,y,w,h,theta
#     """
#     area1 = boxes1[2] * boxes1[3]
#     area2 = boxes2[2] * boxes2[3]
#     r1 = ((boxes1[0], boxes1[1]), (boxes1[2], boxes1[3]), boxes1[4])
#     r2 = ((boxes2[0], boxes2[1]), (boxes2[2], boxes2[3]), boxes2[4])
#     int_pts = cv2.rotatedRectangleIntersection(r1, r2)[1]
#     if int_pts is not None:
#         order_pts = cv2.convexHull(int_pts, returnPoints=True)
#         int_area = cv2.contourArea(order_pts)
#         ious = int_area * 1.0 / (area1 + area2 - int_area)
#     else:
#         ious = 0
#     return ious

--------------------------------------------

正常运行啦~撒花