Releases: james77777778/keras-image-models
Releases · james77777778/keras-image-models
0.2.5
What's Changed
- Apply a more defensive way to fix redownloading issue by @james77777778 in #55
Full Changelog: 0.2.4...0.2.5
0.2.4
Major Features and Improvements
- Add more docstrings
- Fix redownloading issue when using
keras.saving.load_model
What's Changed
- Add docstrings and expose some useful properties for all models by @james77777778 in #48
- Add docstrings to
kimm.list_models
,kimm.utils.get_reparameterized_model
by @james77777778 in #49 - Add docstrings for
kimm.blocks.*
by @james77777778 in #50 - Add docstrings for
kimm.timm_utils.*
by @james77777778 in #51 - Fix redownloading issue by @james77777778 in #54
Full Changelog: 0.2.2...0.2.4
0.2.2
Major Features and Improvements
- Add
GhostNetV3
(currently onlyGhostNetV3W100
has ImageNet weights) - Improve codes
What's Changed
- Refactor reparameterizable layers and add
GhostNetV3
by @james77777778 in #47
Full Changelog: 0.2.1...0.2.2
0.2.1
Major Features and Improvements
Add arbitrary shape support for MobileViT*
and VisionTransformer*
import numpy as np
import kimm
x = np.random.uniform(size=[1, 150, 150, 3])
model = kimm.models.MobileViTXXS(input_shape=[150, 150, 3])
print(model.predict(x, verbose=0).shape) # [1, 1000]
model = kimm.models.VisionTransformerTiny16(input_shape=[150, 150, 3])
print(model.predict(x, verbose=0).shape) # [1, 1000]
Add more model weights
kimm.models.ConvNeXtXLarge
kimm.models.VisionTransformerLarge16
kimm.models.VisionTransformerLarge32
Code cleanup and minor improvements
What's Changed
- Add arbitrary shape support for ViT series by @james77777778 in #46
Full Changelog: 0.2.0...0.2.1
0.2.0
Major Features and Improvements
- Improve UX of APIs
What's Changed
- Refactor model definition by @james77777778 in #42
- Refactor APIs using
namex
by @james77777778 in #43
Full Changelog: 0.1.8...0.2.0
0.1.8
Major Features and Improvements
- Add
HGNet
andHGNetV2
What's Changed
- Update
pyproject.toml
by @james77777778 in #37 - Update
README.md
by @james77777778 in #38 - Update
requirements.txt
andpyproject.toml
by @james77777778 in #40 - Add
HGNet
andHGNetV2
by @james77777778 in #41
Full Changelog: 0.1.7...0.1.8
0.1.7
Major Features and Improvements
- Add
MobileOne
What's Changed
- Bump the github-actions group with 2 updates by @dependabot in #34
- Add
MobileOne
by @james77777778 in #36
New Contributors
- @dependabot made their first contribution in #34
Full Changelog: 0.1.6...0.1.7
0.1.6
Major Features and Improvements
- Add support for reparameterization technique
- Add
RepVGG
What's Changed
- Add
RepVGG
by @james77777778 in #32
Full Changelog: 0.1.5...0.1.6
0.1.5
Major Features and Improvements
- Add
InceptionNeXt
- Add
MobileViTV2
What's Changed
- Add
InceptionNeXt
andMobileViTV2
and change naming ofMobileNet*V2
andMobileNet*V3
by @james77777778 in #30
Full Changelog: 0.1.4...0.1.5
0.1.4
Major Features and Improvements
- Add
export_tflite
andexport_onnx
APIs - Add support of
channels_first
data format
What's Changed
- Add
export_tflite
andexport_onnx
and supportchannels_first
by @james77777778 in #28 - Improve
export.*
APIs by @james77777778 in #29
Full Changelog: 0.1.3...0.1.4