Development of an economical miniaturized platform for monitoring inherent biophysical properties of milk is imperative for tamper-proof milk adulteration detection. Towards this, herein, we demonstrate synthesis and evaluation of a paper-based scalable pH sensor derived from electrospun halochromic nanofibers. The sensor manifests into three unique color-signatures corresponding to pure (6.6 ≤ pH ≤ 6.9), acidic (pH < 6.6), and basic (pH > 6.9) milk samples, enabling a colorimetric detection mechanism. In a practical prototype, color transitions on the sensor strips are captured using smartphone camera and subsequently assigned to one of the three pH ranges using an image-based classifier. Specifically, we implemented three well-known machine learning algorithms and compared their classification performances. For a standard training-to-test ratio of 80:20, support vector machines achieved nearly perfect classification with average accuracy of 99.71%. © 2018, Springer Science+Business Media, LLC, part of Springer Nature.