Abstract:
Defect detection in fabric production lines is a crucial and indispensable task to
maintain quality in the textile industry. The current manual annotation scheme causes the
fabric industry considerable losses. In order to address this issue a real-time detection system
is proposed in this thesis. The system is based on a SOTA deep-learning detection algorithm
which is optimized to achieve real-time performance. The detection architecture is trained on
a self-gathered SOTA dataset from an industrial environment. Deploying the trained model
on an actual real-time operating compactor resulted in 89% accuracy when evaluated
