You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@inproceedings{chaudhary2019ritnet,
title={RITnet: real-time semantic segmentation of the eye for gaze tracking},
author={Chaudhary, Aayush K and Kothari, Rakshit and Acharya, Manoj and Dangi, Shusil and Nair, Nitinraj and Bailey, Reynold and Kanan, Christopher and Diaz, Gabriel and Pelz, Jeff B},
booktitle={2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW)},
pages={3698--3702},
year={2019},
organization={IEEE}
}
Requirements:
Anaconda
Conda environment with Python 3.8
Results:
This is the result trained from 1 epoch of 8000 objects (~34,000 objects in the training set). These are trained with 1 GPU NVIDIA GTX 1060 Max-Q 6 Gb. Trained with normal CE loss.
Input:
Inference
This is the result trained from 10 epoch of 34,000 objects per epoch. These are trained with 1 GPU NVIDIA RTX 3080 12 Gb. Trained with normal CE loss
History
Input
Inference
This is the result trained from 10 epoch of 34,000 objects per epoch. These are trained with 1 GPU NVIDIA RTX 3080 12 Gb. Trained with normal CE and GDL loss.
History
Input
Inference
This is the result trained from 10 epoch of 34,000 objects per epoch. These are trained with 1 GPU NVIDIA RTX 3080 12 Gb. Trained with normal CE, GDL loss, and Surface Loss.
History
Input
Inference
Surface Loss: This is the example result of generating distance matrix based on preprocessed label.
Input (not preprocessed label)
Distance matrix
BAL Loss:
This is the example result of canny edge detection based on preprocessed label.
Input (not preprocessed label)
Canny
Model with BAL Loss integration is still in debugging stage and has not been trained.
