RootRotAI: A Deep Learning-Based Mobile and Web Application for Dry Root Rot Disease Identification and Assessment in Chickpea Under Research Environment

Dry Root Rot (DRR), caused by the soil-borne fungal pathogen Macrophomina phaseolina, is an emerging threat to global chickpea production. Screening large chickpea germplasm collections, mutant populations, and transgenic lines for DRR resistance remains challenging due to manual disease scoring, which is subjective, labor-intensive and often prone to rating bias. To address this, our laboratory has previously developed RootRotAI 2.0, a set of deep learning models developed using TensorFlow to support automated detection and assessment of DRR from images of chickpea collected under laboratory conditions. In the current study, we focus on the high-throughput deployment of these models by developing a mobile application, RootRotAI 2.2, and, a web application, RootRotAI 2.1, to facilitate a user-friendly interface and broader accessibility. These applications allow users to upload or capture images to automatically diagnose and assess DRR severity. RootRotAI 2.2 and 2.1 integrate two transformer-based multitask deep learning models now optimized and deployed via TensorFlow Lite for efficient on-device inference. These tools can be accessed via https://github.com/scipdatabase/RootRotAI and mobile app is available in play store. The application supports three imaging modalities, a handheld camera, a root scanner, and a light microscope each with modality-specific backgrounds. These tools offer a real-time assessment capabilities through an offline mobile mode and a high-capacity platform for large-scale data analysis via the online web application. While image-based disease detection tools are common, RootRotAI serves as a novel and key digital resource for the identification of DRR in chickpea, ultimately facilitating crop improvement programs and contributes to sustainable agriculture.