Algorithm 1 Proposed Glaucoma detection and classification framework.

Input: Image sequences x with three class labels where x ∈ t (where t = 1,2,3).

Outputs: Predicted glaucoma detection for each image sequence and classification for each image.

1. Divide CNN network into two parts: Detection-Net CNN for glaucoma detection and Classification-Net CNN for glaucoma classification estimation.

2. Partition data into training and test sets.

3. Form a pool of apex features based on the training set for Detection-Net.

4. Part 1: Detection-Net

5. if detect Normal

6. stop

7. else glaucoma

8. end if

9. Part 2: Classification-Net

10. for each glaucoma image sequence do

11. for x = 1 to t do

12. Train a Classification-Net CNN for each glaucoma class.

13. end for

14. end for

15. for each glaucoma test sequence do

16. Obtain predicted glaucoma disease levels (Advance, Early, and Moderate).

17. Construct an array of glaucoma disease levels.

18. end for