PUPIL/IRIS/SCLERA DETECTOR (YOLOV4)
Brief Introduction
This model expects a cropped image of an eye as input.
It returns with the sclera, iris and pupil bounding box.
This model can be used for both colour and infrared images.
I created this model back in 2021, so it's not a cutting-edge solution. Also, because it's an older model, it's large in size and not suitable for running a demo in a browser, so there isn't online demo available for it.
However, there is a desktop demo of the solution that can be run on Windows, which is available here:
To be able to run the demo you need to install the latest version of the Microsoft Visual C++ Redistributable from the official Microsoft website.
Business case
Localisation of the iris, can be used to implement biometric identification systems.
This model also lets you measure pupil dilation, which can be used to improve emotion recognition or measure cognitive workload.
Number of classes
3 class: (sclera, iris, pupil)
Metrics
detections_count = 63837, unique_truth_count = 31416
class_id = 0, name = sclera, ap = 99.95% (TP = 10102, FP = 4478)
class_id = 1, name = iris, ap = 95.31% (TP = 10421, FP = 10445)
class_id = 2, name = pupil, ap = 60.28% (TP = 10270, FP = 9892)
for conf_thresh = 0.25, precision = 0.55, recall = 0.98, F1-score = 0.71
for conf_thresh = 0.25, TP = 30793, FP = 24815, FN = 623, average IoU = 48.97 %
IoU threshold = 50 %, used Area-Under-Curve for each unique Recall
mean average precision (mAP@0.50) = 0.851784, or 85.18 %
License
By purchasing or downloading any project, you agree to the full license terms, which you can view here: LINK