I gave 10 classes (Tea, coffee, boost, horlicks, milo etc.,) in training data, but the result is showing me only one class during detection with accuracy of 70-80%.
(I am using SSD model and detectnet)
Hi @VK01, how many items are in your training dataset, and are they evenly distributed across the classes?
If you are using the latest jetson-inference, during training you can run train_ssd.py with --validation-mean-ap and after each epoch it will print out the accuracy of each class. That way you can tell if the model is properly trained or not.
1 Like
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.