International Journal of Agriculture Innovation, Technology and Globalisation (4 papers in press)

Regular Issues

• Deployable lightweight deep learning models for multiclass classification of cocoa and coffee seeds  
  by Emmanuel Asante, Ezekiel Mensah Martey, Eric Opoku, Obed Appiah, Emmanuel Effah 
  Abstract: Accurate cocoa and coffee seed classification is vital for quality control, especially in sub-Saharan Africa, where manual inspection is subjective and labour-intensive. This study tests deep learning models for automated seed classification using multi-source images from variable field conditions. Twelve CNNs, including high-capacity and lightweight architectures, were trained and evaluated following a unified protocol. All models were exported to TensorFlow Lite, quantised to 8-bit, and benchmarked on deployment-ready graphs. Evaluation across accuracy, precision, recall, F1-score, size, parameters, FLOPs, and inference latency showed that lightweight models achieved the best trade-off between accuracy and latency. MobileNetV3Small had the highest performance at 98.7% accuracy and 79 ms latency, followed by MNasNet (96%, 71 ms). DenseNet121 performed well (95.8%) but with a higher computational cost. These findings show compact neural networks, when optimised, can match or outperform larger models for real-time mobile deployment.
  Keywords: multiclass cocoa-coffee classification; deployable deep learning models; lightweight CNNs; agricultural innovation; seed quality assessment.
  DOI: 10.1504/IJAITG.2026.10078417
   
  
• AeroAgriNet: swarm intelligent flying edge machines for precision Agriculture 4.0  
  by Mohammad Shahnawaz Shaikh 
  Abstract: This study presents AeroAgriNet, a swarm-intelligent flying-edge framework designed to enable autonomous, low-latency aerial intelligence for precision agriculture in connectivity-constrained farmlands. The architecture allows each UAV to operate as a mobile edge node with on-board AI inference, decentralised coordination, IoT-supported decision-making, and federated learning. Experimental and simulation results demonstrate significant performance improvements over cloud-based systems, achieving latency of 94
  Keywords: computational offloading; distributed inference; edge computing; federated learning; federated learning; FL; multi UAV collaboration; real-time analytics; swarm robotics; wireless sensor networks; SDG 2 zero hunger; SDG 9 industry; innovation and infrastructure.
  DOI: 10.1504/IJAITG.2026.10078988
   
  
• Building sustainable food systems: the role of agricultural innovation in urbanising Nigeria  
  by Joel Tobiloba Adeyemo, Johnson Olusola Olasebikan 
  Abstract: As Nigerias cities continue to swell and its farmlands strain under the weight of rural exodus, a pressing question looms: can innovation keep food on the table in a rapidly urbanising nation? This study examines the interplay between urbanisation, agricultural innovation, and food security from 1980 to 2022. Using the autoregressive distributed lag (ARDL) model, results show that urbanisation significantly undermines food security, while agricultural innovation has a negative effect when considered in isolation. However, their interaction yields a positive and significant impact, indicating that innovation can mitigate the food security pressures of urban growth when strategically integrated. To ensure the reliability of these findings, a series of robustness checks was conducted, including out-of-sample forecast evaluation, estimation using the dynamic ordinary least squares (DOLS) technique, and causality analyses through Granger and wavelet coherence tests, which capture the time-frequency relationships among the variables of interest. Policy recommendations emphasise coordinated planning that links innovation with infrastructural, educational, and institutional reforms, complemented by coherent trade and urban policies to foster a resilient food system.
  Keywords: food security; urbanisation; agricultural innovation; autoregressive distributed lag; ARDL; dynamic ordinary least squares; DOLS; wavelet coherence.
  DOI: 10.1504/IJAITG.2026.10079589
   
  
• A study on the daily food purchasing strategies of married women in the context of pesticide residue food safety risks  
  by Qi-Wei Cheng, Shu-Mei Wang, Pei-Chang Wen 
  Abstract: This study explores how five married women responsible for household grocery shopping form and implement food purchasing strategies in response to food safety concerns, particularly pesticide residues. Using qualitative in-depth interviews, it draws on Bourdieus lifestyle theory and consumer culture perspectives. The findings show that women adopt diverse strategies, such as buying from trusted vendors, relying on organic or traceability labels, maintaining long-term ties with sellers, and using informal networks to assess safety. These strategies are influenced by caregiving responsibilities, life stage and health experiences, price, convenience, and food safety incidents. Participants continually balance cost, convenience, and risk, sometimes opting for avoidance or home preparation to minimise risks. Beyond economic reasoning, practices are shaped by embodied experiences and cultural values. Recommendations include clearer labelling of high-risk items, strengthening retailer knowledge, and embedding risk communication in everyday contexts.
  Keywords: pesticide residues; food safety risks; married women; lifestyle theory; consumer culture theory.
  DOI: 10.1504/IJAITG.2026.10079994