The Artificial Intelligence of Things project began as a cat‑versus‑dog classification task, testing different AI paradigms in resource‑constrained environments. Early models struggled, with spiking neural networks performing below expectations and accuracy plateauing around 70%. Persistence alone wasn’t enough—limited datasets and hyperparameter tuning proved insufficient. The breakthrough came when the team applied transfer learning, leveraging pre‑trained models to push accuracy above 90%. Achieving this required not only technical creativity but also strong collaboration. Interdisciplinary and intercultural differences demanded adaptability, open communication, and mutual respect. By sharing insights and analyzing each other’s code, the team overcame obstacles together. This experience taught them resilience, problem‑solving, and the power of teamwork in transforming setbacks into impactful research outcomes.
