Computer Vision News - November 2022

41 Glaucoma and Segmentation And the DRISHTI edgesc = color.gray2rgb(1.0*edgesc) ax.imshow(img*edgesc) ax.set_title('Correct') ax = fig.add_subplot(1, 3, 3) (13, 128, 128, 3) {"model_id":"84b48c7de6f44158b8b43299c6ecf555","version_major":2,"version_minor":0} Dice mean= 0.90240574 Std= 0.07061479 best= 0.9616514 worst= 0.67507553 IOU mean= 0.8287271 Std= 0.10222741 best= 0.92613536 worst= 0.50952005 image #11 0.9616514 image #7 0.67507553 Conclusion This is an amazing result but of course, there’s room for improvement. I intentionally change the last part of the network, to let you be able to have space to experiment and make the network better (both in UNet terms but also to achieve higher accuracy). What do you think? I would be very happy to see any images of yours using the same dataset. We’ll be back in two months for another coding article and a small challenge (it seems you do enjoy those). Until then, keep experimenting and be curious! 