Skip to content

Latest commit

 

History

History
157 lines (120 loc) · 12 KB

README.md

File metadata and controls

157 lines (120 loc) · 12 KB
Raw

Image Classification Sample

This sample demonstrates a DL model compression in case of the Image Classification problem. The sample consists of basic steps such as DL model initialization, dataset preparation, training loop over epochs and validation steps. The sample receives a configuration file where the training schedule, hyper-parameters, and compression settings are defined.

Features

  • Models form the tf.keras.applications module (ResNets, MobileNets, Inception, etc.) and datasets (ImageNet, CIFAR 10, CIFAR 100) support.
  • Configuration file examples for sparsity, quantization, filter pruning and quantization with sparsity.
  • Export to Frozen Graph or TensorFlow SavedModel that is supported by the OpenVINO™ toolkit.
  • Distributed training on multiple GPUs on one machine is supported using tf.distribute.MirroredStrategy.

Installation

At this point it is assumed that you have already installed nncf. You can find information on downloading nncf here.

To work with the sample you should install the corresponding Python package dependencies:

pip install -r examples/tensorflow/requirements.txt

Quantize Pretrained Model

This scenario demonstrates quantization with fine-tuning of MobileNetV2 on the ImageNet dataset.

Dataset Preparation

The classification sample supports TensorFlow Datasets (TFDS) and TFRecords. The dataset type is specified in the configuration file by setting the "dataset_type" parameter to "tfds" or "tfrecords" accordingly. TFDS is used by default in all provided configuration files.

Using TFDS

Please read the following guide for more information on how to use TFDS to download and prepare a dataset.

For the ImageNet dataset, TFDS requires a manual download. Please refer to the TFDS ImageNet Readme for download instructions. The TFDS ImageNet dataset should be specified in the configuration file as follows:

    "dataset": "imagenet2012",
    "dataset_type": "tfds"

Legacy TFRecords

To download the ImageNet dataset and convert it to TFRecord format, refer to the following tutorial. The ImageNet dataset in TFRecords format should be specified in the configuration file as follows:

    "dataset": "imagenet2012",
    "dataset_type": "tfrecords"

Run Classification Sample

  • If you did not install the package, add the repository root folder to the PYTHONPATH environment variable.
  • Go to the examples/tensorflow/classification folder.

Test Pretrained Model

Before compressing a model, it is highly recommended checking the accuracy of the pretrained model. All models which are supported in the sample has pretrained weights for ImageNet.

To load pretrained weights into a model and then evaluate the accuracy of that model, make sure that the pretrained=True option is set in the configuration file and use the following command:

python main.py \
--mode=test \
--config=configs/quantization/mobilenet_v2_imagenet_int8.json \
--data=<path_to_imagenet_dataset> \
--disable-compression

Compress Pretrained Model

Run the following command to start compression with fine-tuning on all available GPUs on the machine:

python main.py \
--mode=train \
--config=configs/quantization/mobilenet_v2_imagenet_int8.json \
--data=<path_to_imagenet_dataset> \
--log-dir=../../results/quantization/mobilenet_v2_int8

It may take a few epochs to get the baseline accuracy results.

Use the --resume flag with the path to the checkpoint to resume training from the defined checkpoint or folder with checkpoints to resume training from the last checkpoint.

Validate Your Model Checkpoint

To estimate the test scores of your trained model checkpoint, use the following command:

python main.py \
--mode=test \
--config=configs/quantization/mobilenet_v2_imagenet_int8.json \
--data=<path_to_imagenet_dataset> \
--resume=<path_to_trained_model_checkpoint>

Export Compressed Model

To export trained model to the Frozen Graph, use the following command:

python main.py \
--mode=export \
--config=configs/quantization/mobilenet_v2_imagenet_int8.json \
--resume=<path_to_trained_model_checkpoint> \
--to-frozen-graph=../../results/mobilenet_v2_int8.pb

To export trained model to the SavedModel, use the following command:

python main.py \
--mode=export \
--config=configs/quantization/mobilenet_v2_imagenet_int8.json \
--resume=<path_to_trained_model_checkpoint> \
--to-saved-model=../../results/saved_model

To export trained model to the Keras H5, use the following command:

python main.py \
--mode=export \
--config=configs/quantization/mobilenet_v2_imagenet_int8.json \
--resume=<path_to_trained_model_checkpoint> \
--to-h5=../../results/mobilenet_v2_int8.h5

Export to OpenVINO™ Intermediate Representation (IR)

To export a model to the OpenVINO IR and run it using the Intel® Deep Learning Deployment Toolkit, refer to this tutorial.

Results

Model Compression algorithm Dataset Accuracy (Drop) % NNCF config file TensorFlow checkpoint
Inception V3 INT8 (per-tensor, symmetric for weights; per-tensor, symmetric for activations) ImageNet 78.36 (-0.44) inception_v3_imagenet_int8.json Link
Inception V3 Sparsity 54% (Magnitude) ImageNet 77.87 (0.03) inception_v3_imagenet_magnitude_sparsity.json Link
Inception V3 INT8 (per-tensor, symmetric for weights; per-tensor, symmetric for activations) + Sparsity 61% (RB) ImageNet 77.58 (0.32) inception_v3_imagenet_rb_sparsity_int8.json Link
MobileNet V2 INT8 (per-tensor, symmetric for weights; per-tensor, symmetric for activations) ImageNet 71.66 (0.19) mobilenet_v2_imagenet_int8.json Link
MobileNet V2 Sparsity 50% (RB) ImageNet 71.34 (0.51) mobilenet_v2_imagenet_rb_sparsity.json Link
MobileNet V2 int8(per-tensor, symmetric for weights; per-tensor, symmetric for activations) + sparsity 52% (RB) ImageNet 71.0 (0.85) mobilenet_v2_imagenet_rb_sparsity_int8.json Link
MobileNet V3 small INT8 (per-channel, symmetric for weights; per-tensor, asymmetric for activations) ImageNet 67.75 (0.63) mobilenet_v3_small_imagenet_int8.json Link
MobileNet V3 small INT8 (per-channel, symmetric for weights; per-tensor, asymmetric for activations) + Sparsity 42% (RB) ImageNet 67.55 (0.83) mobilenet_v3_small_imagenet_rb_sparsity_int8.json Link
MobileNet V3 large INT8 (per-channel, symmetric for weights; per-tensor, asymmetric for activations) ImageNet 75.02 (0.79) mobilenet_v3_large_imagenet_int8.json Link
MobileNet V3 large INT8 (per-channel, symmetric for weights; per-tensor, asymmetric for activations) + Sparsity 42% (RB) ImageNet 75.28 (0.53) mobilenet_v3_large_imagenet_rb_sparsity_int8.json Link
ResNet50 INT8 (per-tensor, symmetric for weights; per-tensor, symmetric for activations) ImageNet 75.0 (0.04) resnet50_imagenet_int8.json Link
ResNet50 Sparsity 80% (RB) ImageNet 74.36 (0.68) resnet50_imagenet_rb_sparsity.json Link
ResNet50 INT8 (per-tensor, symmetric for weights; per-tensor, symmetric for activations) + Sparsity 65% (RB) ImageNet 74.3 (0.74) resnet50_imagenet_rb_sparsity_int8.json Link
TensorFlow Hub MobileNet V2 Sparsity 35% (Magnitude) ImageNet 71.90 (-0.06) mobilenet_v2_hub_imagenet_magnitude_sparsity.json Link

Results for filter pruning

Model Compression algorithm Dataset Accuracy (Drop) % GFLOPS MParams NNCF config file TensorFlow checkpoint
ResNet50 None ImageNet 75.04 7.75 (100%) 25.5 (100%) - -
ResNet50 Filter Pruning 40%, geometric_median criterion ImageNet 74.98 (0.06) 4.29 (55.35%) 15.8 (61.96%) Link Link
ResNet50 Filter Pruning 40%, geometric_median criterion + INT8 (per-tensor, symmetric for weights; per-tensor, symmetric for activations) ImageNet 75.08 (-0.04) 4.27 (55.10%) 15.8 (61.96%) Link Link

Results for accuracy-aware compressed training

Model Compression algorithm Dataset Accuracy (Drop) % NNCF config file
ResNet50 Sparsity 65% (magnitude) ImageNet 74.37 (0.67) resnet50_imagenet_magnitude_sparsity_accuracy_aware.json