Weed classification is a fundamental perception task for agricultural robots and an essential enabler of precision and sustainable farming. Existing solutions often rely on high-power edge computing platforms, which limit long-term autonomous operation in Internet of Things (IoT) environments. Meanwhile, the computational complexity of high-accuracy deep learning (DL) models hinders their deployment on resource-constrained microcontrollers (MCUs), a critical component of IoT edge nodes. To address these challenges, this article proposes a tiny machine learning (TinyML)-enabled energy-efficient IoT framework for on-device weed classification, integrating a novel 3-D alignment knowledge distillation (TDA-KD) strategy with a lightweight multilayer dilated-convolution student network. The framework enhances knowledge transfer by jointly aligning: 1) individual predictions; 2) inter-sample correlations; and 3) class semantics, further strengthened through a multitemperature calibration mechanism. Experimental results on the DeepWeeds and 4-Weeds datasets demonstrate that the proposed student model achieves over 95% classification accuracy with only 240 K parameters and 87.51 MFLOPs. The model is successfully deployed on an OpenMV H7 Plus board with an STM32H7 MCU, requiring just 105.68 KB Flash memory and achieving an inference time of 378.3 ms with 510.7 mJ energy consumption per sample. A runtime analysis on the Vitirover horticultural robot shows that, compared with a Jetson Nano-based implementation, the proposed IoT pipeline extends operational time by approximately 30.5%. These results highlight the feasibility of deploying high-accuracy weed classification directly on ultralow-power IoT devices, thereby significantly enhancing the autonomy, energy efficiency, and scalability of agricultural robots.

Monitoring livestock feeding behavior may help assess animal welfare and nutritional status, and to optimize pasture management. The need for continuous and sustained monitoring requires the use of automatic techniques based on the acquisition and analysis of sensor data. This work describes an open dataset of acoustic recordings of the foraging behavior of dairy cows. The dataset includes 708 h of daily records obtained using unobtrusive and non-invasive instrumentation mounted on five lactating multiparous Holstein cows continuously monitored for six non-consecutive days in pasture and barn. Labeled recordings precisely delimiting grazing and rumination bouts are provided for a total of 392 h and for over 6,200 ingestive and rumination jaw movements. Companion information on the audio recording quality and expert-generated labels is also provided to facilitate data interpretation and analysis. This comprehensive dataset is a useful resource for studies aimed at exploring new tools and solutions for precision livestock farming.