Use existing voc data to train on persons and background only

[Steffgroe]

Hi, I am playing around for some time with this implementation of faster rcnn. I succeeded to train on the INRIA persons data set myself, however I still can't make this implementation classify persons and background using the pascal voc data. I made the following files for this: When I try to execute the code after adding the data set to the script I get the following error: Preparing training data... Traceback (most recent call last): File "./tools/trainval_net.py", line 105, in imdb, roidb = combined_roidb(args.imdb_name) File "./tools/trainval_net.py", line 76, in combined_roidb roidbs = [get_roidb(s) for s in imdb_names.split('+')] File "./tools/trainval_net.py", line 73, in get_roidb roidb = get_training_roidb(imdb) File tf-faster-rcnn-master/tools/../lib/model/train_val.py", line 332, in get_training_roidb rdl_roidb.prepare_roidb(imdb) File "tf-faster-rcnn-master/tools/../lib/roi_data_layer/roidb.py", line 49, in prepare_roidb assert all(max_classes[nonzero_inds] != 0) AssertionError Command exited with non-zero status 1

How can I change the code to make faster rcnn only detect the person class using the pascal voc data set provided? pascal_voc_person.zip

[Steffgroe]

Fixed by the following code: ` def _load_pascal_annotation(self, index): """ Load image and bounding boxes info from XML file in the PASCAL VOC format. """ filename = os.path.join(self._data_path, 'Annotations', index + '.xml') tree = ET.parse(filename) objs = tree.findall('object') if not self.config['use_diff']:

Exclude the samples labeled as difficult

  non_diff_objs = [
    obj for obj in objs if int(obj.find('difficult').text) == 0]
  # if len(non_diff_objs) != len(objs):
  #     print 'Removed {} difficult objects'.format(
  #         len(objs) - len(non_diff_objs))
  objs = non_diff_objs

cls_objs = [obj for obj in objs if obj.find('name').text in self._classes]
objs = cls_objs
num_objs = len(objs)

boxes = np.zeros((num_objs, 4), dtype=np.uint16)
gt_classes = np.zeros((num_objs), dtype=np.int32)
overlaps = np.zeros((num_objs, self.num_classes), dtype=np.float32)
# "Seg" area for pascal is just the box area
seg_areas = np.zeros((num_objs), dtype=np.float32)

# Load object bounding boxes into a data frame.
for ix, obj in enumerate(objs):
  bbox = obj.find('bndbox')
  # Make pixel indexes 0-based
  x1 = float(bbox.find('xmin').text) - 1
  y1 = float(bbox.find('ymin').text) - 1
  x2 = float(bbox.find('xmax').text) - 1
  y2 = float(bbox.find('ymax').text) - 1
  cls = self._class_to_ind[obj.find('name').text.lower().strip()]
  boxes[ix, :] = [x1, y1, x2, y2]
  gt_classes[ix] = cls
  overlaps[ix, cls] = 1.0
  seg_areas[ix] = (x2 - x1 + 1) * (y2 - y1 + 1)

overlaps = scipy.sparse.csr_matrix(overlaps)

return {'boxes': boxes,
        'gt_classes': gt_classes,
        'gt_overlaps': overlaps,
        'flipped': False,
        'seg_areas': seg_areas}`

[ngthanhvinh]

Hey @Steffgroe , I want to ask whether you succeeded to detect only 'person' and 'background' based on the existing VOC data. I fixed my code as your code above and finished my training. However, in demo the model detected cats and dogs as 'person' with very high accuracy. Do you know why?