KerasHub: Pretrained Models
Getting started Developer guides API documentation Modeling API Model Architectures Tokenizers Preprocessing Layers Modeling Layers Samplers Metrics Pretrained models list
► KerasHub: Pretrained Models / API documentation / Model Architectures / D-FINE / DFineObjectDetectorPreprocessor layer

DFineObjectDetectorPreprocessor layer

[source]

DFineObjectDetectorPreprocessor class

keras_hub.models.DFineObjectDetectorPreprocessor(image_converter=None, **kwargs)

Base class for object detector preprocessing layers.

ObjectDetectorPreprocessor tasks wraps a keras_hub.layers.Preprocessor to create a preprocessing layer for object detection tasks. It is intended to be paired with a keras_hub.models.ImageObjectDetector task.

All ObjectDetectorPreprocessor take three inputs, x, y, and sample_weight. x, the first input, should always be included. It can be a image or batch of images. See examples below. y and sample_weight are optional inputs that will be passed through unaltered. Usually, y will be the a dict of `{"boxes": Tensor(batch_size, num_boxes, 4), "classes": (batch_size, num_boxes)}.

The layer will returns either x, an (x, y) tuple if labels were provided, or an (x, y, sample_weight) tuple if labels and sample weight were provided. x will be the input images after all model preprocessing has been applied.

All ObjectDetectorPreprocessor tasks include a from_preset() constructor which can be used to load a pre-trained config and vocabularies. You can call the from_preset() constructor directly on this base class, in which case the correct class for your model will be automatically instantiated.

Arguments

  • image_converter: Preprocessing pipeline for images.

Examples.

preprocessor = keras_hub.models.ObjectDetectorPreprocessor.from_preset(
    "retinanet_resnet50",
)


----

<span style="float:right;">[[source]](https://github.com/keras-team/keras-hub/tree/v0.23.0/keras_hub/src/models/preprocessor.py#L132)</span>

### `from_preset` method


```python
DFineObjectDetectorPreprocessor.from_preset(
    preset, config_file="preprocessor.json", **kwargs
)

Instantiate a keras_hub.models.Preprocessor from a model preset.

A preset is a directory of configs, weights and other file assets used to save and load a pre-trained model. The preset can be passed as one of:

  1. a built-in preset identifier like 'bert_base_en'
  2. a Kaggle Models handle like 'kaggle://user/bert/keras/bert_base_en'
  3. a Hugging Face handle like 'hf://user/bert_base_en'
  4. a path to a local preset directory like './bert_base_en'

For any Preprocessor subclass, you can run cls.presets.keys() to list all built-in presets available on the class.

As there are usually multiple preprocessing classes for a given model, this method should be called on a specific subclass like keras_hub.models.BertTextClassifierPreprocessor.from_preset().

Arguments

  • preset: string. A built-in preset identifier, a Kaggle Models handle, a Hugging Face handle, or a path to a local directory.

Examples

# Load a preprocessor for Gemma generation.
preprocessor = keras_hub.models.CausalLMPreprocessor.from_preset(
    "gemma_2b_en",
)

# Load a preprocessor for Bert classification.
preprocessor = keras_hub.models.TextClassifierPreprocessor.from_preset(
    "bert_base_en",
)
Preset Parameters Description
dfine_nano_coco 3.79M D-FINE Nano model, the smallest variant in the family, pretrained on the COCO dataset. Ideal for applications where computational resources are limited.
dfine_small_coco 10.33M D-FINE Small model pretrained on the COCO dataset. Offers a balance between performance and computational efficiency.
dfine_small_obj2coco 10.33M D-FINE Small model first pretrained on Objects365 and then fine-tuned on COCO, combining broad feature learning with benchmark-specific adaptation.
dfine_small_obj365 10.62M D-FINE Small model pretrained on the large-scale Objects365 dataset, enhancing its ability to recognize a wider variety of objects.
dfine_medium_coco 19.62M D-FINE Medium model pretrained on the COCO dataset. A solid baseline with strong performance for general-purpose object detection.
dfine_medium_obj2coco 19.62M D-FINE Medium model using a two-stage training process: pretraining on Objects365 followed by fine-tuning on COCO.
dfine_medium_obj365 19.99M D-FINE Medium model pretrained on the Objects365 dataset. Benefits from a larger and more diverse pretraining corpus.
dfine_large_coco 31.34M D-FINE Large model pretrained on the COCO dataset. Provides high accuracy and is suitable for more demanding tasks.
dfine_large_obj2coco_e25 31.34M D-FINE Large model pretrained on Objects365 and then fine-tuned on COCO for 25 epochs. A high-performance model with specialized tuning.
dfine_large_obj365 31.86M D-FINE Large model pretrained on the Objects365 dataset for improved generalization and performance on diverse object categories.
dfine_xlarge_coco 62.83M D-FINE X-Large model, the largest COCO-pretrained variant, designed for state-of-the-art performance where accuracy is the top priority.
dfine_xlarge_obj2coco 62.83M D-FINE X-Large model, pretrained on Objects365 and fine-tuned on COCO, representing the most powerful model in this series for COCO-style tasks.
dfine_xlarge_obj365 63.35M D-FINE X-Large model pretrained on the Objects365 dataset, offering maximum performance by leveraging a vast number of object categories during pretraining.

image_converter property

keras_hub.models.DFineObjectDetectorPreprocessor.image_converter

The image converter used to preprocess image data.

DFineObjectDetectorPreprocessor layer
DFineObjectDetectorPreprocessor class
Load a preprocessor for Gemma generation.
Load a preprocessor for Bert classification.