YAML Metadata Warning: The pipeline tag "gaze-estimation" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, image-text-to-image, image-text-to-video, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other

EyeGaze: Optimized for Mobile Deployment

Eye gaze estimation from cropped eye images

Predicts gaze direction (pitch, yaw) from 96x160 grayscale eye images using the EyeNet model.

This model is an implementation of EyeGaze found here.

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

Model Details

Model Type: Model_use_case.gaze_estimation
Model Stats:
- Model checkpoint: checkpoint.pt
- Input resolution: 96x160
- Number of parameters: 2.58M
- Model size (float): 9.6MB
- Model size (w8a16): 3.3 MB

Model	Precision	Device	Chipset	Target Runtime	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit	Target Model
EyeGaze	float	QCS8275 (Proxy)	Qualcomm® QCS8275 (Proxy)	TFLITE	30.523 ms	3 - 12 MB	CPU	EyeGaze.tflite
EyeGaze	float	QCS8450 (Proxy)	Qualcomm® QCS8450 (Proxy)	TFLITE	21.249 ms	3 - 26 MB	CPU	EyeGaze.tflite
EyeGaze	float	QCS8550 (Proxy)	Qualcomm® QCS8550 (Proxy)	TFLITE	18.818 ms	3 - 6 MB	CPU	EyeGaze.tflite
EyeGaze	float	QCS8550 (Proxy)	Qualcomm® QCS8550 (Proxy)	ONNX	27.586 ms	49 - 52 MB	CPU	EyeGaze.onnx.zip
EyeGaze	float	QCS9075 (Proxy)	Qualcomm® QCS9075 (Proxy)	TFLITE	26.755 ms	3 - 13 MB	CPU	EyeGaze.tflite
EyeGaze	float	SA7255P ADP	Qualcomm® SA7255P	TFLITE	30.523 ms	3 - 12 MB	CPU	EyeGaze.tflite
EyeGaze	float	SA8255 (Proxy)	Qualcomm® SA8255P (Proxy)	TFLITE	18.761 ms	0 - 3 MB	CPU	EyeGaze.tflite
EyeGaze	float	SA8295P ADP	Qualcomm® SA8295P	TFLITE	17.314 ms	3 - 19 MB	CPU	EyeGaze.tflite
EyeGaze	float	SA8650 (Proxy)	Qualcomm® SA8650P (Proxy)	TFLITE	20.287 ms	3 - 5 MB	CPU	EyeGaze.tflite
EyeGaze	float	SA8775P ADP	Qualcomm® SA8775P	TFLITE	26.755 ms	3 - 13 MB	CPU	EyeGaze.tflite
EyeGaze	float	Samsung Galaxy S24	Snapdragon® 8 Gen 3 Mobile	TFLITE	16.779 ms	3 - 24 MB	CPU	EyeGaze.tflite
EyeGaze	float	Samsung Galaxy S24	Snapdragon® 8 Gen 3 Mobile	ONNX	22.182 ms	50 - 71 MB	CPU	EyeGaze.onnx.zip
EyeGaze	float	Samsung Galaxy S25	Snapdragon® 8 Elite For Galaxy Mobile	TFLITE	14.284 ms	3 - 18 MB	CPU	EyeGaze.tflite
EyeGaze	float	Samsung Galaxy S25	Snapdragon® 8 Elite For Galaxy Mobile	ONNX	20.737 ms	52 - 64 MB	CPU	EyeGaze.onnx.zip
EyeGaze	float	Snapdragon 8 Elite Gen 5 QRD	Snapdragon® 8 Elite Gen5 Mobile	TFLITE	11.528 ms	3 - 14 MB	CPU	EyeGaze.tflite
EyeGaze	float	Snapdragon 8 Elite Gen 5 QRD	Snapdragon® 8 Elite Gen5 Mobile	ONNX	20.425 ms	50 - 62 MB	CPU	EyeGaze.onnx.zip
EyeGaze	float	Snapdragon X Elite CRD	Snapdragon® X Elite	ONNX	9.91 ms	67 - 67 MB	CPU	EyeGaze.onnx.zip
EyeGaze	w8a16	QCS8550 (Proxy)	Qualcomm® QCS8550 (Proxy)	ONNX	37.434 ms	61 - 74 MB	CPU	EyeGaze.onnx.zip
EyeGaze	w8a16	RB3 Gen 2 (Proxy)	Qualcomm® QCS6490 (Proxy)	ONNX	67.761 ms	52 - 66 MB	CPU	EyeGaze.onnx.zip
EyeGaze	w8a16	RB5 (Proxy)	Qualcomm® QCS8250 (Proxy)	ONNX	53.172 ms	61 - 74 MB	CPU	EyeGaze.onnx.zip
EyeGaze	w8a16	Samsung Galaxy S24	Snapdragon® 8 Gen 3 Mobile	ONNX	31.932 ms	70 - 96 MB	CPU	EyeGaze.onnx.zip
EyeGaze	w8a16	Samsung Galaxy S25	Snapdragon® 8 Elite For Galaxy Mobile	ONNX	27.864 ms	69 - 88 MB	CPU	EyeGaze.onnx.zip
EyeGaze	w8a16	Snapdragon 7 Gen 4 QRD	Snapdragon® 7 Gen 4 Mobile	ONNX	59.669 ms	70 - 90 MB	CPU	EyeGaze.onnx.zip
EyeGaze	w8a16	Snapdragon 8 Elite Gen 5 QRD	Snapdragon® 8 Elite Gen5 Mobile	ONNX	27.041 ms	68 - 83 MB	CPU	EyeGaze.onnx.zip
EyeGaze	w8a16	Snapdragon X Elite CRD	Snapdragon® X Elite	ONNX	18.388 ms	100 - 100 MB	CPU	EyeGaze.onnx.zip

Installation

Install the package via pip:

pip install qai-hub-models

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

Sign-in to Qualcomm® AI Hub Workbench 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.eyegaze.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.eyegaze.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.eyegaze.export

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.eyegaze import Model

# Load the model
torch_model = Model.from_pretrained()

# Device
device = hub.Device("Samsung Galaxy S25")

# 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 Workbench. Sign up for access.

Run demo on a cloud-hosted device

You can also run the demo on-device.

python -m qai_hub_models.models.eyegaze.demo --eval-mode on-device

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.eyegaze.demo -- --eval-mode on-device

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 EyeGaze's performance across various devices here. Explore all available models on Qualcomm® AI Hub

License

The license for the original implementation of EyeGaze can be found here.
The license for the compiled assets for on-device deployment can be found here

References

Source Model Implementation

Community

Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
For questions or feedback please reach out to us.

Downloads last month: 10

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support