Detecting more than 2 digits

[lelmedda]

Hi, I was wondering if this code can detect more than 2 digits per picture. I am training it with pictures with 4 digits, and it has only been detecting 2.

[penny4860]

I don't think training has been done properly. Can you show the detected image?

[lelmedda]

The solution I have found for that so far is to divide the image in two parts each with 2 digits since 2 are detected easily.

[lelmedda]

Do you know happen to remember what file I could use to print the digits that have been detected ?

[penny4860]

I didn't understand the question.

[lelmedda]

I just want to save the detected digits to a database, but I am not sure in what file I should be adding the code for that. It's probably obvious, but I haven't worked with CNN before.

[Kamleshsam]

It creates various bounding boxes instead of 2 for 2 digit number . Is this happening for everyone?

[Laleh4]

Hi,
I have this problem, please !!
RuntimeError: The layer has never been called and thus has no defined output shape.
It happens in feature.py !!!!!!!

RuntimeError Traceback (most recent call last)
~\AppData\Local\Temp/ipykernel_1144/3722384473.py in
2
3 # 1. create yolo instance
----> 4 yolo_detector = create_yolo("ResNet50", ["0", "1", "2", "3", "4", "5", "6", "7", "8", "9"], 416)

~\Documents\Pythons\test_yolo\Penny4860\Yolo-digit-detector\yolo\frontend.py in create_yolo(architecture, labels, input_size, anchors, coord_scale, class_scale, object_scale, no_object_scale)
33 n_classes = len(labels)
34 n_boxes = int(len(anchors)/2)
---> 35 yolo_network = create_yolo_network(architecture, input_size, n_classes, n_boxes)
36 yolo_loss = YoloLoss(yolo_network.get_grid_size(),
37 n_classes,

~\Documents\Pythons\test_yolo\Penny4860\Yolo-digit-detector\yolo\backend\network.py in create_yolo_network(architecture, input_size, nb_classes, nb_box)
14 nb_box):
15 feature_extractor = create_feature_extractor(architecture, input_size)
---> 16 yolo_net = YoloNetwork(feature_extractor,
17 input_size,
18 nb_classes,

~\Documents\Pythons\test_yolo\Penny4860\Yolo-digit-detector\yolo\backend\network.py in init(self, feature_extractor, input_size, nb_classes, nb_box)
30
31 # 1. create full network
---> 32 grid_size = feature_extractor.get_output_size()
33
34 # make the object detection layer

~\Documents\Pythons\test_yolo\Penny4860\Yolo-digit-detector\yolo\backend\utils\feature.py in get_output_size(self)
57
58 def get_output_size(self):
---> 59 output_shape = self.feature_extractor.get_output_shape_at(-1)
60 assert output_shape[1] == output_shape[2]
61 return output_shape[1]

~\AppData\Roaming\Python\Python38\site-packages\keras\engine\base_layer.py in get_output_shape_at(self, node_index)
1962 RuntimeError: If called in Eager mode.
1963 """
-> 1964 return self._get_node_attribute_at_index(node_index, 'output_shapes',
1965 'output shape')
1966

~\AppData\Roaming\Python\Python38\site-packages\keras\engine\base_layer.py in _get_node_attribute_at_index(self, node_index, attr, attr_name)
2558 """
2559 if not self._inbound_nodes:
-> 2560 raise RuntimeError('The layer has never been called '
2561 'and thus has no defined ' + attr_name + '.')
2562 if not len(self._inbound_nodes) > node_index:

RuntimeError: The layer has never been called and thus has no defined output shape.