LEATHER VISION: AI -BASED SURFACE DEFECT DETECTION IN LEATHER USING DEEP NEURAL INSPECTION SYSTEM

LEATHER VISION: AI -BASED SURFACE DEFECT DETECTION IN LEATHER USING DEEP NEURAL INSPECTION SYSTEM 1Dr. S. SANJEEVA RAO, 2PASHAPU AKHILA, 3ANTHATI SHILPA, 4BADRI SANTHOSH 03 05 The Leather Vision system is an advanced artificialintelligencebased solution designed to automatethe detection of surface defects in leather materialsusing deep learning techniques Traditional leatherinspection methods rely on manual observationwhich is timeconsuming subjective and prone toinconsistencies due to human fatigue and varyingexpertise This project introduces a computervisiondriven approach that leverages convolutionalneural networks CNNs to enhance accuracy andreliability in defect detection Highresolutionimages of leather surfaces are captured undercontrolled lighting conditions and undergopreprocessing steps such as noise removalnormalization and contrast enhancement toimprove feature quality A transfer learningbasedmodel using MobileNetV2 is employed to extractdeep features and classify defects such as cutsscars wrinkles holes and grain irregularities Thesystem is trained on a labeled dataset enabling it todistinguish between defective and nondefectiveregions with high precision Performanceevaluation is conducted using metrics such asaccuracy precision recall and F1scoredemonstrating significant improvement overtraditional inspection methods The proposedsystem supports realtime defect detection makingit suitable for industrial applications where speedand consistency are critical By reducing humandependency and minimizing material wastage thesystem contributes to improved quality control andcost efficiency in leather manufacturing Overallthis project highlights the transformative potentialof artificial intelligence in modernizing industrialinspection systems

2026