CLI for image processing with histograms, binary treshold and other functions

Banks lose customers silently. Without a model, the retention team has no idea who is about to leave until it is too late. Goal: flag at-risk customers early enough to act. Built a full ML pipeline on 10K customers. Result: 83% Recall, meaning 8 out of 10 churners identified before they leave, giving the retention team an actionable list.

A test framework to assert that CSS doesn't exceeds certain thresholds.

Images segmentation by thresholding using Otsu's method, and RGB to intensity conversion using various methods from scratch

Cryptographically secured multi-candidate election system using a treshold version of Paillier Cryptosystem and Zero-Knowledge proofs.

Proyek ini bertujuan untuk melakukan segmentasi area penyakit pada daun menggunakan pendekatan berbasis pengolahan citra digital (Image Processing). Fokus utama adalah untuk meningkatkan akurasi segmentasi dengan beberapa metode preprocessing, seperti CLAHE dan Gaussian Blur, serta evaluasi kinerja model menggunakan metrik IoU dan Dice Score