amdegroot
IndexError: too many indices for array
File "/home/kun/Software/ssd-pytorch-master/data/voc0712.py", line 145, in pull_item img, boxes, labels = self.transform(img, target[:, :4], target[:, 4]) IndexError: too many indices for array
I want to train voc dataset, but after i modify the path and run train.py, the error shows up. Please help me . Thank you .
Hi @mama110,
I was facing this error too, but while training on my own dataset. The problem for me was an annotation file without any object in it.
Don't know if this will help you but, here is the script I used to check my annotations files :
import argparse
import sys
import cv2
import os
import os.path as osp
import numpy as np
if sys.version_info[0] == 2:
import xml.etree.cElementTree as ET
else:
import xml.etree.ElementTree as ET
parser = argparse.ArgumentParser(
description='Single Shot MultiBox Detector Training With Pytorch')
train_set = parser.add_mutually_exclusive_group()
parser.add_argument('--root', help='Dataset root directory path')
args = parser.parse_args()
CLASSES = ( # always index 0
'aeroplane', 'bicycle', 'bird', 'boat',
'bottle', 'bus', 'car', 'cat', 'chair',
'cow', 'diningtable', 'dog', 'horse',
'motorbike', 'person', 'pottedplant',
'sheep', 'sofa', 'train', 'tvmonitor')
annopath = osp.join('%s', 'Annotations', '%s.{}'.format("xml"))
imgpath = osp.join('%s', 'JPEGImages', '%s.{}'.format("jpg"))
def vocChecker(image_id, width, height, keep_difficult = False):
target = ET.parse(annopath % image_id).getroot()
res = []
for obj in target.iter('object'):
difficult = int(obj.find('difficult').text) == 1
if not keep_difficult and difficult:
continue
name = obj.find('name').text.lower().strip()
bbox = obj.find('bndbox')
pts = ['xmin', 'ymin', 'xmax', 'ymax']
bndbox = []
for i, pt in enumerate(pts):
cur_pt = int(bbox.find(pt).text) - 1
# scale height or width
cur_pt = float(cur_pt) / width if i % 2 == 0 else float(cur_pt) / height
bndbox.append(cur_pt)
print(name)
label_idx = dict(zip(CLASSES, range(len(CLASSES))))[name]
bndbox.append(label_idx)
res += [bndbox] # [xmin, ymin, xmax, ymax, label_ind]
# img_id = target.find('filename').text[:-4]
print(res)
try :
print(np.array(res)[:,4])
print(np.array(res)[:,:4])
except IndexError:
print("\nINDEX ERROR HERE !\n")
exit(0)
return res # [[xmin, ymin, xmax, ymax, label_ind], ... ]
if __name__ == '__main__' :
i = 0
for name in sorted(os.listdir(osp.join(args.root,'Annotations'))):
# as we have only one annotations file per image
i += 1
img = cv2.imread(imgpath % (args.root,name.split('.')[0]))
height, width, channels = img.shape
print("path : {}".format(annopath % (args.root,name.split('.')[0])))
res = vocChecker((args.root, name.split('.')[0]), height, width)
print("Total of annotations : {}".format(i))
@Eralaf It worked like a charm. Thanks a lot!
FYI, you forgot to call vocChecker in main()
I pasted the following code after height, width, channels = img.shape:
res = vocChecker((args.root, name.split('.')[0]), height, width)
facing same error, trying to train my own dataset and annotations are same as coco annotations json files... how to solve the error
img, boxes, labels = self.transform(img, target[:, :4], target[:, 4]) IndexError: too many indices for array
@Eralaf i have tried your method on pycharm,but error shows up :
Traceback (most recent call last):
File "C:/Users/Administrator/Desktop/ssd.pytorch-master/error.py", line 76, in
TypeError: expected str, bytes or os.PathLike object, not NoneType.
so could tell me how to solve the problem ,thanks! (my torch version is 1.0)
@barcahoo Did you call the script with the root argument ?
Let's say the script I used to check my annotations files is called check.py :
python check.py --root="Dataset root directory path"
In this root directory there must be an "Annotations" folder and a "JPEGImages" folder.
If you don't want to call the script like this you can replace the args.root by a variable of your choice like myPath="Dataset root directory path";)
@charan1561 it's a bit late, sorry. When you say your annotations are same as coco annotations, do you mean your dataset is a subset from the cocodataset or your annotations files have the same structure ? Did you try checking your annotations in case there are empty ones (without objects) ? If so, the idea is to not use them :)
@Eralaf thanks! it works, but what puzzies me is that, I added a print log to the top of this sentence, like
print("target {}".format(target))
img, boxes, labels = self.transform(img, target[:, :4], target[:, 4])
why didn't it print out the target of empty, instead, it printed a target with shape (1, 5)
I have this error too，and i use @Eralaf method to check xml
finally，i Found the cause of the mistake
if not self.keep_difficult and difficult: continue
Original code dose not keep difficulte instances in training make xml parse have no object of labels and boxes，so run to
img, boxes, labels = self.transform(img, target[:, :4], target[:, 4])
have IndexError: too many indices for array
I think that will happend when there are not objects in the xml file,because this code does not process this condition
I think that will happend when there are not objects in the xml file,because this code does not process this condition
I feel the same way, but how to solve it ?
I think that will happend when there are not objects in the xml file,because this code does not process this condition
I think that will happend when there are not objects in the xml file,because this code does not process this condition
I feel the same way, but how to solve it ?
I think that will happend when there are not objects in the xml file,because this code does not process this condition
Hi @mama110,
I was facing this error too, but while training on my own dataset. The problem for me was an annotation file without any object in it.
Don't know if this will help you but, here is the script I used to check my annotations files :
import argparse import sys import cv2 import os import os.path as osp import numpy as np if sys.version_info[0] == 2: import xml.etree.cElementTree as ET else: import xml.etree.ElementTree as ET parser = argparse.ArgumentParser( description='Single Shot MultiBox Detector Training With Pytorch') train_set = parser.add_mutually_exclusive_group() parser.add_argument('--root', help='Dataset root directory path') args = parser.parse_args() CLASSES = ( # always index 0 'aeroplane', 'bicycle', 'bird', 'boat', 'bottle', 'bus', 'car', 'cat', 'chair', 'cow', 'diningtable', 'dog', 'horse', 'motorbike', 'person', 'pottedplant', 'sheep', 'sofa', 'train', 'tvmonitor') annopath = osp.join('%s', 'Annotations', '%s.{}'.format("xml")) imgpath = osp.join('%s', 'JPEGImages', '%s.{}'.format("jpg")) def vocChecker(image_id, width, height, keep_difficult = False): target = ET.parse(annopath % image_id).getroot() res = [] for obj in target.iter('object'): difficult = int(obj.find('difficult').text) == 1 if not keep_difficult and difficult: continue name = obj.find('name').text.lower().strip() bbox = obj.find('bndbox') pts = ['xmin', 'ymin', 'xmax', 'ymax'] bndbox = [] for i, pt in enumerate(pts): cur_pt = int(bbox.find(pt).text) - 1 # scale height or width cur_pt = float(cur_pt) / width if i % 2 == 0 else float(cur_pt) / height bndbox.append(cur_pt) print(name) label_idx = dict(zip(CLASSES, range(len(CLASSES))))[name] bndbox.append(label_idx) res += [bndbox] # [xmin, ymin, xmax, ymax, label_ind] # img_id = target.find('filename').text[:-4] print(res) try : print(np.array(res)[:,4]) print(np.array(res)[:,:4]) except IndexError: print("\nINDEX ERROR HERE !\n") exit(0) return res # [[xmin, ymin, xmax, ymax, label_ind], ... ] if __name__ == '__main__' : i = 0 for name in sorted(os.listdir(osp.join(args.root,'Annotations'))): # as we have only one annotations file per image i += 1 img = cv2.imread(imgpath % (args.root,name.split('.')[0])) height, width, channels = img.shape print("path : {}".format(annopath % (args.root,name.split('.')[0]))) res = vocChecker((args.root, name.split('.')[0]), height, width) print("Total of annotations : {}".format(i))
Excuse me , When we check out these files that are empty, what do we do with them? delete?Hope to receive our reply !
From the beginning to the end of the loss has always been equal to 2.000, what is the reason
