mobilenetv2-10

<!--- SPDX-License-Identifier: Apache-2.0 -->

MobileNet

Use cases

MobileNet models perform image classification - they take images as input and classify the major object in the image into a set of pre-defined classes. They are trained on ImageNet dataset which contains images from 1000 classes. MobileNet models are also very efficient in terms of speed and size and hence are ideal for embedded and mobile applications.

Description

MobileNet improves the state-of-the-art performance of mobile models on multiple tasks and benchmarks as well as across a spectrum of different model sizes. MobileNet is based on an inverted residual structure where the shortcut connections are between the thin bottleneck layers. The intermediate expansion layer uses lightweight depthwise convolutions to filter features as a source of non-linearity. Additionally, it removes non-linearities in the narrow layers in order to maintain representational power.

Model

MobileNet reduces the dimensionality of a layer thus reducing the dimensionality of the operating space. The trade off between computation and accuracy is exploited in Mobilenet via a width multiplier parameter approach which allows one to reduce the dimensionality of the activation space until the manifold of interest spans this entire space. The below model is using multiplier value as 1.0.

• Version 2:

Compared with the fp32 MobileNet v2-1.0, int8 MobileNet v2-1.0's Top-1 accuracy decline ratio is 1.70%, Top-5 accuracy decline ratio is 0.92% and performance improvement is 1.05x.

Note the performance depends on the test hardware.

Performance data here is collected with Intel® Xeon® Platinum 8280 Processor, 1s 4c per instance, CentOS Linux 8.3, data batch size is 1.

Inference

We used MXNet as framework with gluon APIs to perform inference. View the notebook imagenet_inference to understand how to use above models for doing inference. Make sure to specify the appropriate model name in the notebook.

Input

All pre-trained models expect input images normalized in the same way, i.e. mini-batches of 3-channel RGB images of shape (N x 3 x H x W), where N is the batch size, and H and W are expected to be at least 224. The inference was done using jpeg image.

Preprocessing

The images have to be loaded in to a range of [0, 1] and then normalized using mean = [0.485, 0.456, 0.406] and std = [0.229, 0.224, 0.225]. The transformation should preferrably happen at preprocessing. Check imagenet_preprocess.py for code.

Output

The model outputs image scores for each of the 1000 classes of ImageNet.

Postprocessing

The post-processing involves calculating the softmax probablility scores for each class and sorting them to report the most probable classes. Check imagenet_postprocess.py for code.

To do quick inference with the model, check out Model Server.

Dataset

Dataset used for train and validation: ImageNet (ILSVRC2012). Check imagenet_prep for guidelines on preparing the dataset.

Validation accuracy

The accuracies obtained by the model on the validation set are mentioned above. The accuracies have been calculated on center cropped images with a maximum deviation of 1% (top-1 accuracy) from the paper.

Training

We used MXNet as framework with gluon APIs to perform training. View the training notebook to understand details for parameters and network for each of the above variants of MobileNet.

Validation

We used MXNet as framework with gluon APIs to perform validation. Use the notebook imagenet_validation to verify the accuracy of the model on the validation set. Make sure to specify the appropriate model name in the notebook.

Quantization

MobileNet v2-1.0-int8 and MobileNet v2-1.0-qdq are obtained by quantizing MobileNet v2-1.0-fp32 model. We use Intel® Neural Compressor with onnxruntime backend to perform quantization. View the instructions to understand how to use Intel® Neural Compressor for quantization.

Environment

onnx: 1.9.0 onnxruntime: 1.8.0

Prepare model

wget https://github.com/onnx/models/raw/main/vision/classification/mobilenet/model/mobilenetv2-12.onnx

Model quantize

Make sure to specify the appropriate dataset path in the configuration file.

bash run_tuning.sh --input_model=path/to/model \  # model path as *.onnx
--config=mobilenetv2.yaml \
--output_model=path/to/save

References

• MobileNet-v2 Model from the paper MobileNetV2: Inverted Residuals and Linear Bottlenecks

• MXNet, Gluon model zoo, GluonCV

• Intel® Neural Compressor

Contributors

• ankkhedia (Amazon AI)
• abhinavs95 (Amazon AI)
• mengniwang95 (Intel)
• airMeng (Intel)
• ftian1 (Intel)
• hshen14 (Intel)

License

Apache 2.0