Internet of Things (IoT) devices often operate with limited resources while interacting with users and their environment, generating a wealth of data. Machine learning models interpret such sensor data, enabling accurate predictions and informed decisions. However, the sheer volume of data from billions of devices can overwhelm networks, making traditional cloud data processing inefficient for IoT applications. This paper presents a comprehensive survey of recent advances in models, architectures, hardware, and design requirements for deploying machine learning on low-resource devices at the edge and in cloud networks. Prominent IoT devices tailored to integrate edge intelligence include Raspberry Pi, NVIDIA’s Jetson, Arduino Nano 33 BLE Sense, STM32 Microcontrollers, SparkFun Edge, Google Coral Dev Board, and Beaglebone AI. These devices are boosted with custom AI frameworks, such as TensorFlow Lite, OpenEI, Core ML, Caffe2, and MXNet, to empower ML and DL tasks (e.g., object detection and gesture recognition). Both traditional machine learning (e.g., random forest, logistic regression) and deep learning methods (e.g., ResNet-50, YOLOv4, LSTM) are deployed on devices, distributed edge, and distributed cloud computing. Moreover, we analyzed 1000 recent publications on “ML in IoT” from IEEE Xplore using support vector machine, random forest, and decision tree classifiers to identify emerging topics and application domains. Hot topics included big data, cloud, edge, multimedia, security, privacy, QoS, and activity recognition, while critical domains included industry, healthcare, agriculture, transportation, smart homes and cities, and assisted living. The major challenges hindering the implementation of edge machine learning include encrypting sensitive user data for security and privacy on edge devices, efficiently managing resources of edge nodes through distributed learning architectures, and balancing the energy limitations of edge devices and the energy demands of machine learning.

Improving agricultural productivity is essential due to rapid population growth, making early detection of crop diseases crucial. Although deep learning shows promise in smart agriculture, practical applications for identifying wheat diseases in complex backgrounds are limited. In this paper, we propose CropNet, a hybrid method that utilizes Red, Green, and Blue (RGB) imaging and a transfer learning approach combined with shallow convolutional neural networks (CNN) for further feature refinement. To develop our customized model, we conducted an extensive search for the optimal deep learning architecture. Our approach involves freezing the pre-trained model for feature extraction and adding a custom trainable CNN layer. Unlike traditional transfer learning, which typically uses trainable dense layers, our method integrates a trainable CNN, deepening the architecture. We argue that pre-trained features in transfer learning are better suited for a custom shallow CNN followed by a fully connected layer, rather than being fed directly into fully connected layers. We tested various architectures for pre-trained models including EfficientNetB0 and B2, DenseNet, ResNet50, MobileNetV2, MobileNetV3-Small, and Inceptionv3. Our approach combines the strengths of pre-trained models with the flexibility of custom architecture design, offering efficiency, effective feature extraction, customization options, reduced overfitting, and differential learning rates. It distinguishes itself from classical transfer learning techniques, which typically fine-tune the entire pre-trained network. Our aim is to provide a lightweight model suitable for resource-constrained environments, capable of delivering outstanding results. CropNet achieved 99.80% accuracy in wheat disease detection with reduced training time and computational cost. This efficient performance makes CropNet promising for practical implementation in resource-constrained agricultural settings, benefiting farmers and enhancing production.