public class YoloModelLoader extends ObjectDetectionModelLoader
ModelLoader
for YOLO models.
These models were built as part of the Gluon CV library and imported into DJL.
Yolo is a model to solve Application.CV.OBJECT_DETECTION
. Prior models like
ObjectDetectionModelLoader
were built around classifiers and would classify at various
locations in the image simultaneously. However, this is fairly inefficient. Yolo instead uses a
regression that will predict both the bounding boxes and class probabilities leading to better
performance and better precision, although it can increase localization errors (the boxes are
less accurate). [paper]
YOLO is currently the best object detection model in the DJL Model Zoo in terms of both performance and prediction quality.
factories, modelZoo, resource
Constructor and Description |
---|
YoloModelLoader(Repository repository,
java.lang.String groupId,
java.lang.String artifactId,
java.lang.String version,
ModelZoo modelZoo)
Creates the Model loader from the given repository.
|
loadModel, loadModel, loadModel
createModel, getApplication, getArtifactId, listModels, loadModel, toString
public YoloModelLoader(Repository repository, java.lang.String groupId, java.lang.String artifactId, java.lang.String version, ModelZoo modelZoo)
repository
- the repository to load the model fromgroupId
- the group id of the modelartifactId
- the artifact id of the modelversion
- the version number of the modelmodelZoo
- the modelZoo type that is being used to get supported engine types