search

SUYEgit / Surgery-Robot-Detection-Segmentation

Object detection and segmentation for a surgery robot using Mask-RCNN on Python 3, Keras, and TensorFlow..
Other
181 stars 83 forks source link

no instances to display error #5

Closed bellakate closed 6 years ago

bellakate commented 6 years ago

hello sir.. i trained on multiple classes and the training was successful. when i try to use the model to detect on a single image, it shows no instances to display error. Can you please tell me what causes the error? any help will be greatly appreciated!

SUYEgit commented 6 years ago

Hi, thank you for using the codes.

No instance means that the model you trained can't detect any instance that belongs to classes you defined. I recommend:

  1. Check the training loss to see if your model has converged.
  2. If the loss has already come to a flat area, maybe you can check if your model has already overfit the training set by detecting images in training set.