This work sought to develop an inelastic scattering imaging system based on Raman spectroscopy for the detection of the fungal phytopathogen, Pseudocercospora fijiensis , which causes Black Sigatoka disease in banana crops, very important in Colombian agro-industrial economy. This system consists of a modified stereoscope with an optical setup able to simultaneously capture spectral images together with its Raman spectra. The camera has two different bandpass filters attached, centered in the spectral region of C=O stretching of Chitin and the equatorial bending vibration of beta-1,3-glucan, molecules of the fungal cell wall. In this way, the system can get images with unique spectral features, suitable for training a convolutional neural network in order to get a recognition pattern of the fungal strain growing in the PDA agar. As a result, the instrument was able to detect the presence of P.fijiensis over the culture media.