The effect of length and position of cumin hypocotyl explants on physical and morphological properties of callus were analyzed by machine vision. Explants were selected at three levels (3, 5 and 7 mm) and set on solid media (B5 containing 0.6 mg l−1 NAA and 0.1 mg l−1 BAP) horizontally (Hp) or vertically (Vp). Results showed that mean values of fresh weight (FW) and volume (VC) of calluses in Hp (0.214 g and 0.205 cm3) were bigger than evaluations for Vp (0.194 g and 0.181 cm3); length of explant had a significant effect on major axis length (a), roundness (R), elongation ratio (Er) and rectangularity ratio (Rr) of calluses (P ≤ 0.05). The density of callus for Vp (1.075 gr cm−3) was larger than evaluations for Hp (1.055 gr cm−3) and the ratio of uncovered area to volume (AUC/Vc) was generally influenced by Vp (P ≤ 0.01). Results also showed that the factor of weighted density (WD) had better prediction performance relative to area of callus (AC) for modelling the FW of calluses. Results indicated that the generalized extreme value (GEV) distribution model was the most likely, and the Weibull distribution model was the least likely probability density function model for modeling AC, FW and VC distributions of calluses.