Performance Evaluation of Stainless Steel Plate Defects Using Deep Learning Approach

Over the past decade, the detection and classification of Steel surface defect image has been a great challenge in computational methodology. This research work aims at classify the Steel surface defect image which can be used to assess quality of steel surface and also measure the performance metrics using this computational methodology. The Proposed work based on Fractional Jaya Optimizer-based Deep Convolutional Neural Network (FJO-DCNN). The segments are generated through the clustering mechanism named Particle Swarm Optimization (PSO), which ensure the effectiveness of optimal segment selection that yields to detect Steel surface defect image more accurately. However, the optimal segments are effectively selected that yield to detect the Steel surfacedefect regions. This experimentation is carried out using the NEU-DET database. Finally, the results are carried out by using this hybrid algorithm and attained the better performance value. The proposed work achieves in FJO-DCNN for Steel surface defect image computed the best values for accuracy, sensitivity and specificity respectively.