release v0.0.1

Release Notes - Gaze Estimation Models v0.0.1

We are excited to announce the release of our pre-trained models for gaze estimation. This release includes several state-of-the-art architectures trained over 200 epochs to provide robust and accurate gaze prediction across different use cases.

Included Models

Model	Weights	Size	Epochs	MAE
ResNet-18	resnet18.pt	43 MB	200	12.84
ResNet-34	resnet34.pt	81.6 MB	200	11.33
ResNet-50	resnet50.pt	91.3 MB	200	11.34
MobileNet V2	mobilenetv2.pt	9.59 MB	200	13.07
MobileOne S0	mobileone_s0_fused.pt	4.8 MB	200	*
MobileOne S1	mobileone_s1_fused.pt	8.2 MB	200	*
MobileOne S2	mobileone_s2_fused.pt	14.8 MB	200	*
MobileOne S3	mobileone_s3_fused.pt	23.4 MB	200	*
MobileOne S4	mobileone_s4_fused.pt	34.2 MB	200	*

'*' - will be uploaded in near future.

Key Highlights

• Versatile Architectures: Includes ResNet (18, 34, 50) for robust performance across various scenarios, along with MobileNet V2 for lightweight deployments, and MobileOne series (S0-S4) for cutting-edge efficiency.
• Pre-trained Weights: All models are trained for 200 epochs, ensuring high accuracy and generalization.
• Ready for Integration: Models can be easily integrated into your existing projects for tasks such as gaze direction estimation, facial analysis, and more.

These models are now available for download and use in your gaze estimation projects. We look forward to your feedback and contributions!

Full Changelog: https://github.com/yakhyo/gaze-estimation/commits/v0.0.1