How do I draw multiple bounding boxes while using few shot learning method

Hello,

  I'm trying to practice few shot learning on my own to identify grape bunches. In most of the images there are multiple grape bunches. How do I use this method to identify multiple grape bunches and draw bounding boxes around it in a same image?

Thanks!

Hi @Kirang96!

You mean drawing boxes annotating the images with colab_utils.annotate, or after running inference with the already trained model?

In the first case, if you want to annotate more than 1 element in an image, just draw another bounding box without clicking the submit button, and when you are done, annotating the grape bunches, just clicking submit will count every box you have drown.

The second case is more intuitive, with the function plot_detections you have implemented at the start of the few shot detection colab notebook, just adjust the paremeters that you have to give it as input, and it will draw the bounding boxes correctly.

The parameters for plot_detections are the ones that the neural network outputs while running inference for each test image (the detections map has all the information of the detections and you can acces them this way: detections[‘detection_boxes’]).

I hope this helps you! You can ask anything else you need :slight_smile:
Keep it up!

1 Like