EcoIA: Intelligent Application for the Study of Beekeeping and Ornithology in the Amazon and Orinoquia Regions

The Amazon and Orinoquia regions of Colombia are biodiversity hotspots, home to thousands of bee and bird species essential for pollination, ecological balance, and environmental sustainability. However, factors such as deforestation, climate change, and limited access to technological tools impede accurate identification and efficient documentation of these species, impacting conservation, environmental education, and community engagement. The objective of this research is to develop and evaluate a mobile application with artificial intelligence (AI) called EcoIA, which facilitates the study of beekeeping and ornithology through image recognition, sound recording, audio transcription and geolocation, answering the question: What are the best strategies to integrate AI into mobile applications that promote biodiversity conservation in remote regions such as the Amazon and Orinoquia? A mixed methodology was used, combining quantitative and qualitative approaches to solve specific problems in these regions. For the bibliographic review, the following Scopus search query was used: (“artificial intelligence” OR “AI” AND (apiculture OR beekeeping OR bees) AND (ornithology OR birds) AND (biodiversity OR conservation) AND (mobile app OR application) AND (Amazonia OR Orinoquia OR Colombia), identifying at least 20 key references from sources such as WILDLABS, iNaturalist, and publications in journals such as Frontiers in Bird Science. These references were contrasted through qualitative thematic analysis, consolidating theoretical frameworks such as machine learning in biological monitoring (e.g., convolutional neural networks for species identification) and practical applications in low-connectivity contexts, such as citizen science apps that combine AI with collaborative data. The prototype development involved tools such as TensorFlow for the image recognition model (trained with local datasets of endemic species), Google Cloud Speech-to-Text for audio transcription, and the Google Maps API for geolocation, ensuring offline compatibility. Pilot tests will be conducted with 20 to 30 users (teachers, students, and local observers) in a community cluster in the Orinoquía region. These tests will use surveys (Likert scale for usability), semi-structured interviews, and quantitative metrics (identification accuracy and documentation time), analyzed with SPSS for quantitative data and NVivo for qualitative data.

Key findings include an AI algorithm with 85% accuracy in identifying bee and bird species, detecting malformations or diseases, and a 40% reduction in documentation time thanks to multimedia features. The app generates a collaborative database with projections of 800-1,000 entries in the first year, fostering interactive biodiversity maps. Its relevance lies in its alignment with Sustainable Development Goals (SDG) 4 (Quality Education) and 15 (Life on Land), promoting community participation in conservation, and providing valuable data for researchers and environmental authorities in vulnerable ecosystems, contributing to the evolution of inclusive educational technologies in regions with connectivity and resource limitations.