Yolo
This module belongs to the " Computer Vision" category .
Last updated
This module belongs to the " Computer Vision" category .
Last updated
The Yolo module is used to initiate a yolo (You Only Look Once) object detection model.
The Yolo module is a compact toolset for capturing visual features.
Among others, we can :
select the optimizer : adam,rmsprop, adagrad, adamax ,sgd
set : beta, gradient, epsilon, learning rate, momentum, rho
The Yolo module includes many functionalities we can select either by ticking the check boxes, by selecting them in dropdowns or by setting them manually.
We may for instance check the : data augmentation , trainable skeleton, save checkpoint to local , early stopping. All of them are boolean.
Speaking about anchor boxes, default values are preset. However, in need, we might always add our own values to obtain proper results.
As a features extractor , we have the choice between:
darknet,
mobilenet ,
resnet50/101/152_v2,
nasnet,
and xception.
Finally, labels, non max suppression threshold, number of classes, true boxes,object scale , object threshold and warm Up Batches are all variable parameters to help us fine-tune the Yolo' s accuracy.
To get a better insight on Yolo, check this blog article which explains it in a few lines.
