callback.py question

GroG

Hello,
I have callback.py currently working, but I have 2 questions
1. how do I extract label_str information when there are detections ?
def onNn(nn_packet, decoded_data):
if nn_packet.detections:
for nn_detection in nn_packet.detections:
print(dir(nn_detection))
print(nn_detection.label)
print(nn_detection.confidence)

it appears the label_str is not set in the either the nn_packet or the decoded one
2. Is there any class definitions or examples for useful information from the callback parameters ?

jakaskerl

Hi GroG
Could you paste the code you are using?
I'm not really sure what callback parameters you are talking about.

Thanks,
Jaka

GroG

it was in the initial question.
the callback onNn gets called with decoded_data, this is in callback.py which was already mentioned.
a "label" is a field in decoded_data, 15 is for "person".
Depending on the network used and what it was trained on, determines how the label_str gets changed from 15 to "person". So, I'm guessing there should be a way to get this from the network config. Is there a way ?

jakaskerl

Hi GroG
nn.get_labels() will give you the labels defined in the models config.json.

GroG

How do i get nn from inside of the callback methods ?
in callback.py
def onNn(nn_packet, decoded_data):
# print the label str of the detected object here
print(f'detected a {nn.get_labels()[detected.label]}')

detected.label is 15, for this network "person" is 15, how do i get the mapping ?

jakaskerl

Hi GroG
Well, for NNPacket, it should be label_str. That is, if the mappings are actually defined inside the config file.

Thanks,
Jaka

GroG

The default doesn't have that field, as I mentioned in my very first post.

jakaskerl

GroG
That's why I requested to see your code. Are you using a custom mode? Label string wouldn't be accessible if the mappings aren't defined inside the .json configuration file.

Here is an example: luxonis/depthaiblob/main/depthai_sdk/src/depthai_sdk/nn_models/yolov4_tiny_coco_416x416/config.json

Thanks,
Jaka