Detectron2 model not generating any predictions?

[Tangolin]

I am using this simple piece of code to try out the model

pretrained_model = model_zoo.get('COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml', trained=True)

# for p in pretrained_model.parameters():
#     input(p.shape)

im = Image.open('input.jpg')

t = transforms.ToTensor()
pretrained_model.eval()

with torch.no_grad():
    im = t(im)
    output = pretrained_model([{'image':im, 'height':480, 'width':640}])
    print(output)

However the output I get is

[{'instances': Instances(num_instances=0, image_height=480, image_width=640, fields=[pred_boxes: Boxes(tensor([], device='cuda:0', size=(0, 4))), scores: tensor([], device='cuda:0'), pred_classes: tensor([], device='cuda:0', dtype=torch.int64), pred_masks: tensor([], device='cuda:0', size=(0, 480, 640), dtype=torch.bool)])}]

What may be causing this behaviour? The input image is the one from the colab tutorial notebook so the model should be able to perform on it.