Hugging Face's logo

qualcomm
/
ResNet-Mixed-Convolution

Video Classification
PyTorch
LiteRT
backbone
android
Model card Files Community
ResNet-Mixed-Convolution / README.md
qaihm-bot's picture
qaihm-bot
Upload README.md with huggingface_hub
76be976 verified
preview code
|
raw
history blame
16.1 kB
metadata
library_name: pytorch
license: other
tags:
  - backbone
  - android
pipeline_tag: video-classification

ResNet-Mixed-Convolution: Optimized for Mobile Deployment

Sports and human action recognition in videos

ResNet Mixed Convolutions is a network with a mixture of 2D and 3D convolutions used for video understanding.

This model is an implementation of ResNet-Mixed-Convolution found here.

This repository provides scripts to run ResNet-Mixed-Convolution on Qualcomm® devices. More details on model performance across various devices, can be found here.

Model Details

  • Model Type: Model_use_case.video_classification
  • Model Stats:
    • Model checkpoint: Kinectics-400
    • Input resolution: 112x112
    • Number of parameters: 11.7M
    • Model size (float): 44.7 MB
    • Model size (w8a8): 11.7 MB
Model Precision Device Chipset Target Runtime Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit Target Model
ResNet-Mixed-Convolution float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) TFLITE 343.367 ms 31 - 57 MB NPU ResNet-Mixed-Convolution.tflite
ResNet-Mixed-Convolution float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) QNN 98.835 ms 2 - 12 MB NPU Use Export Script
ResNet-Mixed-Convolution float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) TFLITE 166.491 ms 31 - 69 MB NPU ResNet-Mixed-Convolution.tflite
ResNet-Mixed-Convolution float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) QNN 28.033 ms 2 - 60 MB NPU Use Export Script
ResNet-Mixed-Convolution float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) TFLITE 145.437 ms 31 - 41 MB NPU ResNet-Mixed-Convolution.tflite
ResNet-Mixed-Convolution float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) QNN 14.255 ms 2 - 4 MB NPU Use Export Script
ResNet-Mixed-Convolution float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) TFLITE 155.618 ms 31 - 58 MB NPU ResNet-Mixed-Convolution.tflite
ResNet-Mixed-Convolution float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) QNN 26.1 ms 2 - 17 MB NPU Use Export Script
ResNet-Mixed-Convolution float SA7255P ADP Qualcomm® SA7255P TFLITE 343.367 ms 31 - 57 MB NPU ResNet-Mixed-Convolution.tflite
ResNet-Mixed-Convolution float SA7255P ADP Qualcomm® SA7255P QNN 98.835 ms 2 - 12 MB NPU Use Export Script
ResNet-Mixed-Convolution float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) TFLITE 147.636 ms 31 - 43 MB NPU ResNet-Mixed-Convolution.tflite
ResNet-Mixed-Convolution float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) QNN 14.31 ms 2 - 4 MB NPU Use Export Script
ResNet-Mixed-Convolution float SA8295P ADP Qualcomm® SA8295P TFLITE 166.292 ms 31 - 56 MB NPU ResNet-Mixed-Convolution.tflite
ResNet-Mixed-Convolution float SA8295P ADP Qualcomm® SA8295P QNN 27.631 ms 0 - 18 MB NPU Use Export Script
ResNet-Mixed-Convolution float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) TFLITE 141.0 ms 31 - 42 MB NPU ResNet-Mixed-Convolution.tflite
ResNet-Mixed-Convolution float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) QNN 14.368 ms 3 - 5 MB NPU Use Export Script
ResNet-Mixed-Convolution float SA8775P ADP Qualcomm® SA8775P TFLITE 155.618 ms 31 - 58 MB NPU ResNet-Mixed-Convolution.tflite
ResNet-Mixed-Convolution float SA8775P ADP Qualcomm® SA8775P QNN 26.1 ms 2 - 17 MB NPU Use Export Script
ResNet-Mixed-Convolution float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile TFLITE 143.665 ms 31 - 43 MB NPU ResNet-Mixed-Convolution.tflite
ResNet-Mixed-Convolution float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile QNN 14.222 ms 2 - 25 MB NPU Use Export Script
ResNet-Mixed-Convolution float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 14.349 ms 0 - 73 MB NPU ResNet-Mixed-Convolution.onnx
ResNet-Mixed-Convolution float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile TFLITE 111.595 ms 31 - 65 MB NPU ResNet-Mixed-Convolution.tflite
ResNet-Mixed-Convolution float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile QNN 10.238 ms 2 - 71 MB NPU Use Export Script
ResNet-Mixed-Convolution float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 10.476 ms 2 - 59 MB NPU ResNet-Mixed-Convolution.onnx
ResNet-Mixed-Convolution float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile TFLITE 109.783 ms 29 - 58 MB NPU ResNet-Mixed-Convolution.tflite
ResNet-Mixed-Convolution float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile QNN 10.265 ms 2 - 68 MB NPU Use Export Script
ResNet-Mixed-Convolution float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 10.171 ms 2 - 56 MB NPU ResNet-Mixed-Convolution.onnx
ResNet-Mixed-Convolution float Snapdragon X Elite CRD Snapdragon® X Elite QNN 14.92 ms 2 - 2 MB NPU Use Export Script
ResNet-Mixed-Convolution float Snapdragon X Elite CRD Snapdragon® X Elite ONNX 14.954 ms 22 - 22 MB NPU ResNet-Mixed-Convolution.onnx
ResNet-Mixed-Convolution w8a8 QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) QNN 14.414 ms 1 - 11 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) QNN 6.546 ms 1 - 57 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) QNN 4.171 ms 1 - 3 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) QNN 4.531 ms 1 - 15 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 RB3 Gen 2 (Proxy) Qualcomm® QCS6490 (Proxy) QNN 23.841 ms 1 - 14 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 SA7255P ADP Qualcomm® SA7255P QNN 14.414 ms 1 - 11 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 SA8255 (Proxy) Qualcomm® SA8255P (Proxy) QNN 4.174 ms 1 - 4 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 SA8295P ADP Qualcomm® SA8295P QNN 7.617 ms 1 - 18 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 SA8650 (Proxy) Qualcomm® SA8650P (Proxy) QNN 4.178 ms 1 - 3 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 SA8775P ADP Qualcomm® SA8775P QNN 4.531 ms 1 - 15 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile QNN 4.168 ms 1 - 13 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 4.934 ms 0 - 37 MB NPU ResNet-Mixed-Convolution.onnx
ResNet-Mixed-Convolution w8a8 Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile QNN 3.012 ms 1 - 55 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 3.653 ms 0 - 60 MB NPU ResNet-Mixed-Convolution.onnx
ResNet-Mixed-Convolution w8a8 Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile QNN 2.754 ms 1 - 42 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 3.073 ms 1 - 47 MB NPU ResNet-Mixed-Convolution.onnx
ResNet-Mixed-Convolution w8a8 Snapdragon X Elite CRD Snapdragon® X Elite QNN 4.531 ms 1 - 1 MB NPU Use Export Script
ResNet-Mixed-Convolution w8a8 Snapdragon X Elite CRD Snapdragon® X Elite ONNX 5.278 ms 12 - 12 MB NPU ResNet-Mixed-Convolution.onnx

Installation

Install the package via pip:

pip install "qai-hub-models[resnet-mixed]"

Configure Qualcomm® AI Hub to run this model on a cloud-hosted device

Sign-in to Qualcomm® AI Hub with your Qualcomm® ID. Once signed in navigate to Account -> Settings -> API Token.

With this API token, you can configure your client to run models on the cloud hosted devices.

qai-hub configure --api_token API_TOKEN

Navigate to docs for more information.

Demo off target

The package contains a simple end-to-end demo that downloads pre-trained weights and runs this model on a sample input.

python -m qai_hub_models.models.resnet_mixed.demo

The above demo runs a reference implementation of pre-processing, model inference, and post processing.

NOTE: If you want running in a Jupyter Notebook or Google Colab like environment, please add the following to your cell (instead of the above).

%run -m qai_hub_models.models.resnet_mixed.demo

Run model on a cloud-hosted device

In addition to the demo, you can also run the model on a cloud-hosted Qualcomm® device. This script does the following:

  • Performance check on-device on a cloud-hosted device
  • Downloads compiled assets that can be deployed on-device for Android.
  • Accuracy check between PyTorch and on-device outputs.
python -m qai_hub_models.models.resnet_mixed.export

Profiling Results
------------------------------------------------------------
ResNet-Mixed-Convolution
Device                          : cs_8275 (ANDROID 14)                
Runtime                         : TFLITE                              
Estimated inference time (ms)   : 343.4                               
Estimated peak memory usage (MB): [31, 57]                            
Total # Ops                     : 57                                  
Compute Unit(s)                 : npu (53 ops) gpu (0 ops) cpu (4 ops)

How does this work?

This export script leverages Qualcomm® AI Hub to optimize, validate, and deploy this model on-device. Lets go through each step below in detail:

Step 1: Compile model for on-device deployment

To compile a PyTorch model for on-device deployment, we first trace the model in memory using the jit.trace and then call the submit_compile_job API.

import torch

import qai_hub as hub
from qai_hub_models.models.resnet_mixed import Model

# Load the model
torch_model = Model.from_pretrained()

# Device
device = hub.Device("Samsung Galaxy S24")

# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()

pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])

# Compile model on a specific device
compile_job = hub.submit_compile_job(
    model=pt_model,
    device=device,
    input_specs=torch_model.get_input_spec(),
)

# Get target model to run on-device
target_model = compile_job.get_target_model()

Step 2: Performance profiling on cloud-hosted device

After compiling models from step 1. Models can be profiled model on-device using the target_model. Note that this scripts runs the model on a device automatically provisioned in the cloud. Once the job is submitted, you can navigate to a provided job URL to view a variety of on-device performance metrics.

profile_job = hub.submit_profile_job(
    model=target_model,
    device=device,
)

Step 3: Verify on-device accuracy

To verify the accuracy of the model on-device, you can run on-device inference on sample input data on the same cloud hosted device.

input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
    model=target_model,
    device=device,
    inputs=input_data,
)
    on_device_output = inference_job.download_output_data()

With the output of the model, you can compute like PSNR, relative errors or spot check the output with expected output.

Note: This on-device profiling and inference requires access to Qualcomm® AI Hub. Sign up for access.

Deploying compiled model to Android

The models can be deployed using multiple runtimes:

  • TensorFlow Lite (.tflite export): This tutorial provides a guide to deploy the .tflite model in an Android application.

  • QNN (.so export ): This sample app provides instructions on how to use the .so shared library in an Android application.

View on Qualcomm® AI Hub

Get more details on ResNet-Mixed-Convolution's performance across various devices here. Explore all available models on Qualcomm® AI Hub

License

  • The license for the original implementation of ResNet-Mixed-Convolution can be found here.
  • The license for the compiled assets for on-device deployment can be found here

References

Community