Training a YOLOv3 network on own dataset, saving it and loading it again.

I am working on YOLOv3 example. https://www.mathworks.com/help/vision/ug/object-detection-using-yolo-v3-deep-learning.html#ObjectDetectionUsingYOLOV3DeepLearningExample-7

Given Data:

I have dataset of almost 5000 images. I want to train my network(by giving it 200 images information) at one time. until it reaches a certain point after training where on testing the network it gives desired outputs.

QUESTION:

So, when implementing this, i have to save my network so that i can load it in for next time for further training.

% This is for the time, when i have trained my yolo network for first time.
save('yolov3trainednetworkimg1_220.mat','net');

% Since i have to train again, so here i am loading this file and net will again go for training.
load('yolov3trainednetworkimg1_220.mat')
net=pretrained.net;

Will this work correctly ? For the purpose i want ? Is it possible ?

Answer 1

[Rushil]

Hello

I believe that saving and loading models for further training (or inference) is possible by using “save” and “load” functions.

To know more, follow the link to a previously answered question on MATLAB Answers:

https://mathworks.com/matlabcentral/answers/1850013-how-to-save-a-yolov3-training-to-load-it-again

I hope this helps