Deep Flora: Sistema de Reconhecimento de Imagens de Plantas por meio de Inteligência Artificial

Accurate plant identification is a significant challenge for institutions, researchers, and enthusiasts due to the vast diversity of species and the morphological similarity between many of them. This work presents the development of an intelligent system, based on convolutional neural networks (CNNs), capable of automatically classifying plant images using public databases such as Pl@ntNet-300K and GBIF. The system was implemented in Python, using ResNet-50 and EfficientNet-B3 models fine-tuned with transfer learning, along with a Flask-based web interface to make it accessible to different users. The pipeline encompasses data collection, rigorous preprocessing, and validation to ensure reliable results. Tests in real- world scenarios demonstrated high performance, with the system accurately identifying common species and providing users with complementary information about the analyzed plants.