Deep learning (DL) algorithms have been deployed on an increasing number of end devices to enable various smart applications. Considering the large energy consumption of those algorithms, energy harvesting becomes the most promising energy supply. To adapt to the environmental changes over time and learn new knowledge, frequent over-the-air (OTA) code programming is required to update to the new model that’s incrementally learned on the nearby edge server. However, it is a grand challenge to update the DL code on devices due to the constrained resources and low harvested energy. To address those challenges, this paper proposes a novel intermittent OTA update framework for incremental DL algorithms on energy harvesting IoT devices. Specifically, we propose a delta encoding method to reduce the update code size, a data transmission optimization method to reduce the communication energy consumption, and runtime support to enable efficient intermittent updates. The experimental results demonstrate that the proposed framework can achieve reliable and efficient intermittent updates.