Consuming fresh Leafy Vegetables (LVs) offers numerous health benefits due to their essential vitamins, minerals, and fiber content. However, accurately identifying freshness can be difficult, leading to post-purchase spoilage and waste. This research addresses the challenge of manually classifying the freshness of LVs. The agricultural industry faces numerous challenges throughout the production cycle. When it comes to LVs they have short self-life and their quality deteriorates rapidly over time. Distinguishing between Day01 and Day02 freshness can be particularly tricky, but beyond Day02, wilting accelerates, causing significant losses for consumers, retailers, exporters, and restaurant owners. A model capable of classifying LVs into multiple freshness categories would enable informed inventory management, timely buying and selling decisions, and optimized sales and profits. Fresh LVs were obtained directly from a vendor who supplied directly from farm. Captured images from Day01 to Day03 and saved images under freshness category. This research investigates the use of deep learning pre-trained models using transfer learning for LVs freshness classification. Experimented VGG-16, ResNet50, DenseNet201 architectures. The DenseNet architecture, known for its dense connectivity, feature propagation, reduced parameter count, and strong feature extraction capabilities, was chosen for its effectiveness in vegetable and LV classification. DenseNet with unfrozen denseblock3 and denseblock4 layers, were explored using transfer learning. The proposed DenseNet201 model achieved the highest test accuracy 96.46% with unfreezing denseblock3 and denseblock4. It is accurately classifying the LVs into multiple categories Day01, Day02, Day03.