fizyr / keras-retinanet

Keras implementation of RetinaNet object detection.
Apache License 2.0
4.38k stars 1.96k forks source link

Training images dimesions and effect on output #1568

Closed ShreimHossein closed 2 years ago

ShreimHossein commented 2 years ago

hello, I'd like to know if what I am doing is practically applicable and if I will get results. I am training a model to detect plant diseases from the shapes of the leaves. if the input images are only images of the leaves themselves and small sizes about 250x250, will I be able to get good results if the testing images are of a full plant with multiple leaves and bigger dimensions?
I posted an example of training and testing images.

if its is not the right training set images for the testing images, should I bring images similar to the testing ones (large and with background) and label them myself?

0eacfbb6-3d68-4a89-b8f8-6b2d328102af___RS_HL 7961

download