YOLOR training

Stuff I am working on
Post Reply
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

YOLOR training

Post by Daniel Wee »

Mostly following the steps here:-

https://blog.roboflow.com/train-yolor-o ... m-dataset/

All git clones take place in the Nextcloud/python directory. The training images are placed inside of:-

Code: Select all

Nextcloud/python/yolor/training
Consequently, the Nextcloud/python/yolor/training/yaml.cfg file will need to be edited so that the train and valid paths point to the correct places, namely:-

Code: Select all

../yolor/training/train/images
../yolor/training/valid/images
Daniel Wee
Site Admin
Posts: 2449
Joined: Wed 25 Feb 25 2009 8:00 pm

Re: YOLOR training

Post by Daniel Wee »

At this point, we have two major tasks to complete:-

1. Creating a custom training set using Roboflow, and annotate with the right kinds of classes/categories. This should correspond to the types of fish that we want to identify in the aquarium. So, we need to know:-
a. number of images
b. size of images
c. boundary marking
d. labelling and annotation
e. formatting with Roboflow
f. exporting

2. We will need to automate the detection and generate the output accordingly. This may involve:-
a. running the detection command/process in a python script
b. marking out the boundaries (if necessary) and generating a resultant detection image
c. keeping track of location and count of detections
Post Reply