Is it possible to use 5 channel image in detectron2 deep learning?

[martiInfa]

To run detectron2 for imaging with 5 channels I followed #2062 . I have implement a custom dataloader and changed PIXEL_MEAN/STD but I have this error:
FloatingPointError: Loss became infinite or NaN at iteration=4!
loss_dict = {'loss_cls': 0.5807800889015198, 'loss_box_reg': 0.0, 'loss_mask': 0.0, 'loss_rpn_cls': 6.5862717628479, 'loss_rpn_loc': inf}

This is the implementation of class Trainer(DefaultTrainer)
def custom_mappe(dataset_dict):
dataset_dict = copy.deepcopy(dataset_dict) # it will be modified by code below
# can use other ways to read image
image=read_tiff_with_5_bands(dataset_dict["file_name"])
#image = utils.read_image(dataset_dict["file_name"], format="BGR")
# See "Data Augmentation" tutorial for details usage

#auginput = T.AugInput(image)
#transform = T.Resize((650, 650))(auginput)
#image = torch.from_numpy(auginput.image.transpose(2, 0, 1))
'''
annos = [
    utils.transform_instance_annotations(annotation, [transform], image.shape[1:])
    for annotation in dataset_dict.pop("annotations")
]
'''
annos=[]
for annotation in dataset_dict.pop("annotations"):
    annos+=[annotation]

image = torch.from_numpy(image.transpose(2, 0, 1))
return {
   # create the format that the model expects
    
   "image": image,
   "instances":utils.annotations_to_instances(annos, image.shape[1:]),
    #"istances":dataset_dict['annotations'],
    "width":dataset_dict['width'],
    "height":dataset_dict['height']
}

class Trainer(DefaultTrainer):
@classmethod
def build_train_loader(cls, cfg):
return build_detection_train_loader(cfg, mapper=custom_mappe)