main

fun main()

fun main()

fun main()

This examples demonstrates the inference concept on VGG'16 model and weights loading from KotlinDL txt format:

Weights are loaded from txt files, configuration is loaded from .json file.
Model predicts on a few images located in resources.
Special preprocessing (used in VGG'16 during training on ImageNet dataset) is applied to images before prediction.
No additional training.
No new layers are added.

See also

Detailed description of VGG'16 model and an approach to build it in Keras.

fun main()

This examples demonstrates the inference concept on VGG'16 model and weights loading from outdated or custom weights' schema in .h5 file:

Weights are loaded from .h5 file, configuration is loaded from .json file.
Model predicts on a few images located in resources.
Special preprocessing (used in VGG'16 during training on ImageNet dataset) is applied to images before prediction.
No additional training.
No new layers are added.

NOTE: Also recursivePrintGroupInHDF5File() is helpful to discover hidden schema and paths.

See also

Detailed description of VGG'16 model and an approach to build it in Keras.

fun main()

This examples demonstrates the inference concept on VGG'16 model:

Weights are loaded from .h5 file, configuration is loaded from .json file.
Model predicts on a few images located in resources.
Special preprocessing (used in VGG'16 during training on ImageNet dataset) is applied to images before prediction.
No additional training.
No new layers are added.

See also

Detailed description of VGG'16 model and an approach to build it in Keras.