Skip to content

[ViTDet] Remove hardcoded image preprocessing and add ViTDet ImageConverter #2452

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 2 commits into
base: master
Choose a base branch
from
Open

[ViTDet] Remove hardcoded image preprocessing and add ViTDet ImageConverter #2452

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5d18d9c
Remove hardcoded vitdet image preprocessing and add vitdet converter
DavidLandup0 Oct 29, 2025
0ca60d4
Update keras_hub/src/models/vit_det/vit_det_image_converter.py
DavidLandup0 Oct 29, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions keras_hub/api/layers/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -147,6 +147,9 @@
from keras_hub.src.models.vit.vit_image_converter import (
ViTImageConverter as ViTImageConverter,
)
from keras_hub.src.models.vit_det.vit_det_image_converter import (
ViTDetImageConverter as ViTDetImageConverter,
)
from keras_hub.src.models.whisper.whisper_audio_converter import (
WhisperAudioConverter as WhisperAudioConverter,
)
Expand Down
5 changes: 0 additions & 5 deletions keras_hub/src/models/vit_det/vit_det_backbone.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,4 @@
import keras
from keras import ops

from keras_hub.src.api_export import keras_hub_export
from keras_hub.src.models.backbone import Backbone
Expand Down Expand Up @@ -105,10 +104,6 @@ def __init__(
)
img_size = img_input.shape[-3]
x = img_input
# VITDet scales inputs based on the standard ImageNet mean/stddev.
x = (x - ops.array([0.485, 0.456, 0.406], dtype=x.dtype)) / (
ops.array([0.229, 0.224, 0.225], dtype=x.dtype)
)
x = ViTDetPatchingAndEmbedding(
kernel_size=(patch_size, patch_size),
strides=(patch_size, patch_size),
Expand Down
38 changes: 38 additions & 0 deletions keras_hub/src/models/vit_det/vit_det_image_converter.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,38 @@
from keras_hub.src.api_export import keras_hub_export
from keras_hub.src.layers.preprocessing.image_converter import ImageConverter
from keras_hub.src.models.vit_det.vit_det_backbone import ViTDetBackbone


@keras_hub_export("keras_hub.layers.ViTDetImageConverter")
class ViTDetImageConverter(ImageConverter):
"""Image converter for ViTDet models.

This layer applies ImageNet normalization (mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]) to input images for ViTDet models.

Args:
image_size: int or tuple of (height, width). The output size of the
image. Defaults to `(1024, 1024)`.

Example:
```python
converter = keras_hub.layers.ViTDetImageConverter(image_size=(1024, 1024))
converter(np.random.rand(1, 512, 512, 3)) # Resizes and normalizes
Comment on lines +19 to +20
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Oct 29, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

The example code is a bit misleading and incomplete:

  1. It uses np without showing the import numpy as np statement.
  2. np.random.rand() generates float values in [0, 1). The layer then scales this by 1/255, which is likely not the intended demonstration. Using np.random.randint(0, 256, ...) would better simulate a typical uint8 image for which scale=1.0 / 255.0 is appropriate.

Please add the import inside the ````pythonblock and userandint` for clarity.

Suggested change
converter = keras_hub.layers.ViTDetImageConverter(image_size=(1024, 1024))
converter(np.random.rand(1, 512, 512, 3)) # Resizes and normalizes
converter = keras_hub.layers.ViTDetImageConverter(image_size=(1024, 1024))
converter(np.random.randint(0, 256, size=(1, 512, 512, 3))) # Resizes and normalizes

```
"""

backbone_cls = ViTDetBackbone

def __init__(
self,
image_size=(1024, 1024),
**kwargs,
):
super().__init__(
image_size=image_size,
scale=1.0 / 255.0, # Scale to [0, 1]
offset=None,
norm_mean=[0.485, 0.456, 0.406],
norm_std=[0.229, 0.224, 0.225],
**kwargs,
)
Loading