This repository stores a dataset of high-quality images of stems from 4 sugarcane varieties, obtained under controlled laboratory conditions, built to train Artificial Intelligence systems to perform classification processes. Captured using a full-frame mirrorless camera, the images feature the stems on a neutral black background, with no post-processing applied. The dataset contains images of sugar cane stems captured directly from the camera. It is organized into four compressed files, each representing a specific variety. Inside these files there are JPEG images, approximately 5 megapixels in size, portraying individual or grouped samples. All images were done under the same standardized light conditions.