To interact with an unlabeled rich visual world, the human developing brain learns to differentiate visual events and to generalize across some of them despite their physical variability. Although this perceptual categorization process has been traditionally investigated from a unisensory perspective, the early development of visual categorization is inherently constrained by multisensory inputs. In particular, the visual system being largely immature in infancy, early-maturing sensory systems such as olfaction are ideally suited to support and refine visual development. Odors are relevant cues for young infants providing stability and familiarity within a rapidly changing complex visual environment. Here we provide evidence that perceptual categorization of one of the earliest and most salient visual category for the young infant brain, human faces, is shaped by a familiar odor already learned in utero, the maternal odor. We recorded scalp electroencephalogram (EEG) from 18 four-month-old infants exposed to the maternal or a control odor condition during rapid 6 Hz stream of widely variable natural images, with faces inserted every 6th image (i.e., at 1 Hz). We replicate the infant brain ability to categorize at a glance various faces embedded in a fast train of natural images, with a significant face categorization response at 1 Hz over right posterior cortical regions (de Heering & Rossion, 2015). Strikingly, this visual categorization response is enhanced by concomitant maternal odor inputs, with every single infant brain showing a larger face categorization response in the maternal odor context. The lack of difference between odor conditions for the 6 Hz visual response common to all images excludes a mere enhancement of visual attention in the maternal odor context. Overall, these observations support a multisensory view of category learning and have important implications for our understanding of the development of perceptual categorization in the human brain.