import keras import kimm # List available models print(kimm.list_models("mobileone", weights="imagenet")) # ['MobileOneS0', 'MobileOneS1', 'MobileOneS2', 'MobileOneS3'] # Initialize model with pretrained ImageNet weights # Note: all `kimm` models expect inputs in the value range of [0, 255] by # default if `include_preprocessing=True` x = keras.random.uniform([1, 224, 224, 3]) * 255.0 model = kimm.models.MobileOneS0() y = model.predict(x) print(y.shape) # (1, 1000) # Print some basic information about the model print(model) # <MobileOneS0 name=MobileOneS0, input_shape=(None, None, None, 3), # default_size=224, preprocessing_mode="imagenet", feature_extractor=False, # feature_keys=None> # This information can also be accessed through properties print(model.input_shape, model.default_size, model.preprocessing_mode) # List available feature keys of the model class print(kimm.models.MobileOneS0.available_feature_keys) # ['STEM_S2', 'BLOCK0_S4', 'BLOCK1_S8', 'BLOCK2_S16', 'BLOCK3_S32'] # Enable feature extraction by setting `feature_extractor=True` # `feature_keys` can be optionally specified feature_extractor = kimm.models.MobileOneS0( feature_extractor=True, feature_keys=["BLOCK2_S16", "BLOCK3_S32"] ) features = feature_extractor.predict(x) for feature_name, feature in features.items(): print(feature_name, feature.shape) # BLOCK2_S16 (1, 14, 14, 256), BLOCK3_S32 (1, 7, 7, 1024), ...

import keras import kimm import numpy as np # Initialize a reparameterizable model x = keras.random.uniform([1, 224, 224, 3]) * 255.0 model = kimm.models.MobileOneS0() y = model.predict(x) # Get reparameterized model by kimm.utils.get_reparameterized_model reparameterized_model = kimm.utils.get_reparameterized_model(model) y2 = reparameterized_model.predict(x) np.testing.assert_allclose( keras.ops.convert_to_numpy(y), keras.ops.convert_to_numpy(y2), atol=1e-3 ) # Export model to tflite format kimm.export.export_tflite(reparameterized_model, 224, "model.tflite") # Export model to onnx format # Note: must be "channels_first" format before the exporting # kimm.export.export_onnx(reparameterized_model, 224, "model.onnx")