Abstract—The automatic identification and diagnosis of tomato
leaves diseases are highly desired in field of agriculture information. Recently Deep Convolutional Neural networks (CNN) has
made tremendous advances in many fields, close to computer
vision such as classification, object detection, segmentation,
achieving better accuracy than human-level perception. In spite
of its tremendous advances in computer vision tasks, CNN face
many challenges, such as computational burden and energy,
to be used in mobile phone and embedded systems. In this
study, we propose an efficient smart mobile application model
based on deep CNN to recognize tomato leaf diseases. To build
such application, our model has been inspired from MobileNet
CNN model and can recognize the 10 most common types of
Tomato leaf disease. Trained on tomato leafs dataset, to build
our application 7176 images of tomato leaves are used in the
smart mobile system, to perform a Tomato disease diagnostics.
Index Terms—Convolutional Neural networks, Tomato disease
detection, Smart Mobile application, MobileNet.