InceptionV3 Models

Overview

Inception V3, a convolutional neural network (CNN) from Google’s Inception family, is designed for deep networks with fewer than 25 million parameters. It excels in image analysis and object detection, with applications ranging from computer vision to life sciences like leukemia research. Often, it’s used pre-trained on ImageNet.

Model Conversion Flow

Precondition

Note

For better compatibility, it is recommended to use Python 3.7 when working with these models, as it has higher compatibility with certain libraries and frameworks.

  1. Install Required Libraries:

    Ensure you have the necessary libraries installed:

    pip install torch torchvision

Get Source Model

Follow these steps to use and convert GoogLeNet models using PyTorch and TorchVision. 1. Load and Convert GoogLeNet Model:

Load a pretrained Inception v3 model using PyTorch and TorchVision, create a dummy input tensor for tracing, trace the model to convert it to TorchScript, and finally save the traced model.

Run the following Python script:

python generate_inception_v3_float.py

import torch
import torchvision

model = torchvision.models.inception_v3(pretrained=True)
trace_data = torch.randn(1, 3, 224, 224)
trace_model = torch.jit.trace(model.cpu().eval(), trace_data)
torch.jit.save(trace_model, 'Inception_v3.pt')

Converting Model for Deployment

Before you begin, ensure that the NeuroPilot Converter Tool is installed. If you haven’t installed it yet, please follow the instructions in the “Install and Verify NeuroPilot Converter Tool” section of the same guide.

Quant8 Conversion Process

  1. Generate Calibration Data:

    The following script creates a directory named data and generates 100 batches of random input data, each saved as a .npy file. This data is used for calibration during the quantization process.

    python generate_data_batches.py

    import os
import numpy as np

os.mkdir('data')
for i in range(100):
    data = np.random.randn(1, 3, 224, 224).astype(np.float32)
    np.save('data/batch_{}.npy'.format(i), data)

  2. Convert to Quantized TFLite Format:

    Use the following command to convert the model to a quantized TFLite format using the generated calibration data:

    mtk_pytorch_converter                                 \
    --input_script_module_file=Inception_v3.pt        \
    --output_file=Inception_v3_ptq_quant.tflite       \
    --input_shapes=1,3,224,224                        \
    --quantize=True                                   \
    --input_value_ranges=-1,1                         \
    --calibration_data_dir=data/                      \
    --calibration_data_regexp=batch_.*\.npy

  3. Convert to Quantized DLA Format

Warning

The process of converting this model to DLA format may encounter unsupported operations. Before converting, you need to prune the model using the provided script to ensure compatibility with the DLA converter.

Here is an example of how to prune the Inception_v3_ptq_quant.tflite model using the export_quant_support_op.py script:

import mtk_converter

editor = mtk_converter.TFLiteEditor("Inception_v3_ptq_quant.tflite")
output_file = "Inception_v3_ptq_quant_15-181_sdkv6.tflite"
input_names = ["input.4"]
output_names = ["1707"]
_ = editor.export(output_file=output_file, input_names=input_names, output_names=output_names)

Note

The input_names and output_names specified in this script are based on the example model structure. You need to modify these names according to the input and output tensor names specific to your model. You can inspect the model structure using tools like Netron or TensorFlow utilities to identify the correct tensor names for your model.

Tip

For more detailed information and steps on handling unsupported operations in DLA conversion, please see Unsupported Operations in DLA Conversion (Optional).

  • NeuroPilot SDK tools Download and Convert to DLA:

    1. Download the NeuroPilot SDK All-In-One Bundle:

      Visit the following download page and download the necessary bundle: NeuroPilot Downloads

    2. Extract the Bundle:

      After downloading, extract the bundle using the following command:

      tar zxvf neuropilot-sdk-basic-<version>.tar.gz

    3. Set the Environment Variables:

      Set the environment variables to point to the SDK:

      export LD_LIBRARY_PATH=/path/to/neuropilot-sdk-basic-<version>/neuron_sdk/host/lib

    4. Convert INT8 TFLite Model to DLA Format:

      Use the NeuroPilot Converter Tool to convert your TFLite model into the DLA format. The following example shows how to convert an INT8 TFLite model to DLA format using the specified architecture (mdla3.0).

      /path/to/neuropilot-sdk-basic-<version>/neuron_sdk/host/bin/ncc-tflite --arch=mdla3.0 Inception_v3_ptq_quant_15-181_sdkv6.tflite

Note

To ensure compatibility with your device, please download and use NeuroPilot SDK version 6. Other versions might not be fully supported.

FP32 Conversion Process

  1. Convert to FP32 TFLite Format:

    To convert the model to a non-quantized (FP32) TFLite format, use the following command:

mtk_pytorch_converter                                 \
    --input_script_module_file=Inception_v3.pt        \
    --output_file=Inception_v3.tflite                 \
    --input_shapes=1,3,224,224

  1. Convert to FP32 DLA Format

Warning

The process of converting this model to DLA format may encounter unsupported operations. Before converting, you need to prune the model using the provided script to ensure compatibility with the DLA converter.

Here is an example of how to prune the Inception_v3.tflite model using the export_float_support_op.py script:

import mtk_converter
editor = mtk_converter.TFLiteEditor("Inception_v3.tflite")

output_file = "Inception_v3_15-181_sdkv6.tflite"
input_names = ["input.4"]
output_names = ["1707"]

_ = editor.export(output_file=output_file, input_names=input_names, output_names=output_names)

Note

The input_names and output_names specified in this script are based on the example model structure. You need to modify these names according to the input and output tensor names specific to your model. You can inspect the model structure using tools like Netron or TensorFlow utilities to identify the correct tensor names for your model.

  • Set the Environment and Convert to DLA:

    1. Set the Environment Variables:

      Set the environment variables to point to the SDK:

      export LD_LIBRARY_PATH=/path/to/neuropilot-sdk-basic-<version>/neuron_sdk/host/lib

    2. Convert FP32 TFLite Model to DLA Format:

      Use the NeuroPilot Converter Tool to convert your FP32 TFLite model into the DLA format. The following example shows how to convert an FP32 TFLite model to DLA format using the specified architecture (mdla3.0) and enabling relaxed FP32 operations:

      /path/to/neuropilot-sdk-basic-<version>/neuron_sdk/host/bin/ncc-tflite --arch=mdla3.0 --relax-fp32 Inception_v3_15-181_sdkv6.tflite

Model Information

Note

The models and benchmark data mentioned below have been processed using the mtk_converter.

General Information

The following table contains general information about the model. The details, such as input size, GFLOPS, and number of parameters, are sourced from the official PyTorch documentation at: Inception_V3 Model.

Property

Value

Category

Classification

Input Size

224x224

GFLOPS

1.50

#Params (M)

6.62

Training Framework

PyTorch

Inference Framework

TFLite

Pre-converted Model

Deployable Model

Model Type

Download Link

Supported Backend

Quant8 Model package

Download: Quant8

NeuronSDK

Float32 Model package

Download: Fp32

CPU,GPU,ArmNN,Neuron Stable Delegate,NeuronSDK

Model Properties

  • InceptionV3-quant8

Inputs

Property

Value

Name

x.2

Tensor

int8[1,3,224,224]

Identifier

154

Quantization

Linear

Quantization Range

-1.0039 ≤ 0.0078 * q ≤ 0.9961

Outputs

Property

Value

Name

1707

Tensor

int8[1,1000]

Identifier

73

Quantization

Linear

Quantization Range

-2.0561 ≤ 0.0196 * (q + 23) ≤ 2.9372

  • InceptionV3-fp32

Inputs

Property

Value

Name

x.2

Tensor

float32[1,3,224,224]

Identifier

157

Outputs

Property

Value

Name

1707

Tensor

float32[1,1000]

Identifier

8

Benchmark Results

Note

The benchmark results shown below were measured with performance mode enabled. These numbers are for reference only, as actual performance may vary depending on the hardware and platform used.

Please note the following limitations:

  1. The G350 does not support Neuron Stable Delegate and NeuronSDK because the hardware does not yet support these features.

  2. The model may not run on certain backends due to custom operators generated by the MTK converter. These custom operators are not recognized or supported by the TensorFlow Lite interpreter, which may lead to incompatibility issues during inference.

  3. Running models on the G350 using ArmNN inference may result in a crash due to the model size being too large for the platform to handle.

  • InceptionV3-quant8

Run model (.tflite) 10 times

CPU (Thread:8)

GPU

ARMNN(GpuAcc)

ARMNN(CpuAcc)

Neuron Stable Delegate

NeuronSDK

G350

N/A

N/A

N/A

N/A

N/A

N/A

G510

N/A

N/A

N/A

N/A

N/A

5.59 ms

G700

N/A

N/A

N/A

N/A

N/A

3.04 ms

G1200

N/A

N/A

N/A

N/A

N/A

5.04 ms

  • InceptionV3-fp32

Run model (.tflite) 10 times

CPU (Thread:8)

GPU

ARMNN(GpuAcc)

ARMNN(CpuAcc)

Neuron Stable Delegate

NeuronSDK

G350

405.697 ms (Thread:4)

325.574 ms

275.707 ms

N/A

N/A

N/A

G510

319.684 ms

112.367 ms

98.235 ms

80.029 ms

17.199 ms

17.68 ms

G700

76.034 ms

83.642 ms

70.822 ms

68.439 ms

12.263 ms

12.04 ms

G1200

70.697 ms

68.656 ms

49.117 ms

41.429 ms

11.757 ms

11.04 ms

Run Benchmark Tools

This section will guide you on how to execute the benchmark tool with different delegates and hardware configurations.

  1. First, push your TFLite model to the target device:

adb push <your_tflite_model> /usr/share/label_image/

Make sure to replace <your_tflite_model> with the actual path of your TFLite model.

  1. Next, open an ADB shell to the target device:

adb shell

After this, you can execute the following commands directly from the shell.

Execute on CPU (8 threads)

To execute the benchmark on the CPU using 8 threads, run the following command:

benchmark_model --graph=/usr/share/label_image/<your_tflite_model> --num_threads=8 --num_runs=10

Execute on GPU, with GPU delegate

To execute the benchmark on the GPU using the TensorFlow Lite GPU delegate, run the following command:

benchmark_model --graph=/usr/share/label_image/<your_tflite_model> --use_gpu=1 --allow_fp16=0 --gpu_precision_loss_allowed=0 --num_runs=10

Execute on GPU, with Arm NN delegate

To execute the benchmark on the GPU using the Arm NN delegate, use the following command:

benchmark_model --graph=/usr/share/label_image/<your_tflite_model> --external_delegate_path=/usr/lib/libarmnnDelegate.so.29 --external_delegate_options="backends:GpuAcc" --num_runs=10

Execute on CPU, with Arm NN delegate

To run the benchmark on the CPU using the Arm NN delegate, use the following command:

benchmark_model --graph=/usr/share/label_image/<your_tflite_model> --external_delegate_path=/usr/lib/libarmnnDelegate.so.29 --external_delegate_options="backends:CpuAcc" --num_runs=10

Execute on APU, with Neuron Delegate

For executing on the APU using the Neuron delegate, run the following command:

benchmark_model --stable_delegate_settings_file=/usr/share/label_image/stable_delegate_settings.json --use_nnapi=false --use_xnnpack=false --use_gpu=false --min_secs=20 --graph=/usr/share/label_image/<your_tflite_model>

Note

If you are using the G350 platform, please make the following adjustments:

  • For CPU-based benchmarks, change the –num_threads parameter to 4:

    benchmark_model --graph=/usr/share/label_image/<your_tflite_model> --num_threads=4 --use_xnnpack=0 --num_runs=10

  • For all benchmarks (CPU, GPU, Arm NN), add the parameter –use_xnnpack=0 to disable the XNNPACK delegate

Neuron SDK

Follow these steps to benchmark your TensorFlow Lite model using the Neuron SDK with MDLA 3.0:

  1. Transfer the Model to the Device:

    Use adb to push your TFLite model to the device:

    adb push <your_tflite_model> /user/share/benchmark_dla/

Note

Make sure to push the pruned model (after using the pruning script) to the device to ensure compatibility with the DLA converter. The pruned model should be used instead of the original model for accurate benchmarking.

  1. Access the Device Shell:

    Connect to your device’s shell:

    adb shell

  2. Navigate to the Benchmark Directory:

    Change to the directory where the model is stored:

    cd /user/share/benchmark_dla/

  3. Run the Benchmark:

    Execute the benchmarking script with the following command:

    python3 benchmark.py --file <your_tflite_model> --target mdla3.0 --profile --options='--relax-fp32'

Description:

  • The benchmark.py script runs a performance evaluation on your model using MDLA 3.0.

  • The –file parameter specifies the path to your TFLite model.

  • The –target mdla3.0 option sets the target hardware to MDLA 3.0.

  • The –profile flag enables profiling to provide detailed performance metrics.

  • The –options=’–relax-fp32’ option allows relaxation of floating-point precision to improve compatibility with MDLA.