efficientnet
efficientnet copied to clipboard
qubvel
Implementation of EfficientNet model. Keras and TensorFlow Keras.
> EfficientNet-lite are a set of mobile/IoT friendly image classification models. Notably, while EfficientNet-EdgeTPU that is specialized for Coral EdgeTPU, these EfficientNet-lite models run well on all mobile CPU/GPU/EdgeTPU. >Due...
``` input_image = Input(shape=image_input_shape) eff_model=efn.EfficientNetB3(input_shape=(height, width, 3), include_top=False, weights='noisy-student') model_backbone = Model(eff_model.input,[eff_model.get_layer('block6d_se_squeeze').output,eff_model.output]) timeDistributed_layer = tf.keras.layers.TimeDistributed(model_backbone)(input_image) t = LSTM(128...)(timeDistributed_layer) -->first output of backbone c = Conv2D(128,..)(timeDistributed_layer) --> second output of backbone...
Hi, I am trying to train on a TPU with Kaggle, save the weights, and then predict on a GPU. The problem is that whenever I save my model's weights...
Hi I'm really new to this model Why don't you use Depthwise separable convolution except _depthwise_conv?(`_expand_conv`, `_se_reduce`, `_se_expand`, `_project_conv `) Does it matters for higher acc? model.py ``` python #...
Hi, Is the computation of top1 accuracy and top 5 accuracy is included in the code? In order to reproduce results. Thank YOu
Modify the code to fully support tensorflow2.0. There is no need to install any independent keras library when using tensorflow2.0 and above.
how can i retrian the efficientnet model with tensorflow framework? Can you give me an example?
can we use efficientnet as a feature extractor on high-resolution image input(1024x1024)? It will affect the accuracy or we should expect similar level results? I tried the same but facing...
Hi, there. I found some bug of 'preprocess_input' function when i used the function with tf.data. ## What is this PR? :mag: This issue was occured when we use the...
