Nous avons accueilli ce jeudi 19 septembre, le Professeur Zaid Omar de l’UTM (Universiti Teknologi Malaysia) pour un séminaire du laboratoire.
Title : Is Sustainable Palm Oil Harvesting Viable? An Engineering Solution.
Abstract: The palm oil industry in Malaysia, which generates USD37 billion annually and accounts for 43% of global production, faces significant challenges due to labor-intensive harvesting methods that lead to yield losses from unharvested ripe fruits. To tackle these challenges, researchers at the Universiti Teknologi Malaysia have developed an automated palm fruit ripeness classification system using computer vision techniques. This system employs the YOLOv4 algorithm for fruit localization, a salient segmentation method for background removal, and extracts color and texture features for classification using a Multilayer Perceptron (MLP) classifier. Achieving very promising accuracy, this innovative solution not only improves harvesting efficiency by ensuring that only fully ripened fruits are cut but also enhances sustainability in the palm oil sector, demonstrating the importance of interdisciplinary collaboration in advancing agricultural technologies.