Advancements in artificial intelligence for computer vision and sound detection present significant potential to enhance the analysis of ecological systems. Ecological impact assessments are crucial for sustainable development and wildlife conservation. In this context, we propose an architecture that leverages AI and data integration techniques to monitor birds, aiming to reduce mortality rates among both endangered and non-endangered species while enhancing surveillance in a non-invasive manner. Our architecture combines on-edge devices, including microphones and high-resolution cameras, with public databases to enable real-time bird detection, classification, and integration into live maps. By employing deep learning models such as You Only Look Once (YOLOv11), EfficientNet pretrained networks, Transformers, and self-attention mechanisms, we have developed an efficient system that runs on Nvidia Jetson hardware, connected to the aforementioned on-edge devices. The model was tested in outdoor environments, demonstrating high accuracy, low latency, and stability. The heatmaps produced by our system (utilizing both images and sound) are consistent with other open-source heatmaps, as they display comparable patterns of bird presence in various regions.