Quality Inspection for casting product

Notebook link

https://www.kaggle.com/sonyd4d/qc-using-cnn/notebook

Overview

• Defects are an unwanted thing in casting industry. For removing this defective product all industry have their quality inspection department.
• But,the main problem is this inspection process is carried out manually and it is a very time-consuming process and due to human involvement.
• the results obtained through this method are not 100% accurate.
• This can because of the rejection of the entire order thus creating a loss for company.

Objective

To automate this process using machine learning models

Dataset

• This dataset is of casting manufacturing product.
• Casting is a manufacturing process in which a liquid material is usually poured into a mould, which contains a hollow cavity of the desired shape, and then allowed to solidify.
• Reason for collect this data is casting defects!!
• Casting defect is an undesired irregularity in a metal casting process.
• There are many types of defect in casting like blow holes, pinholes, burr, shrinkage defects, mould material defects, pouring metal defects, metallurgical defects, etc.

EDA

CNN

Model: "sequential"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d (Conv2D)              (None, 150, 150, 32)      320       
_________________________________________________________________
max_pooling2d (MaxPooling2D) (None, 75, 75, 32)        0         
_________________________________________________________________
conv2d_1 (Conv2D)            (None, 38, 38, 64)        18496     
_________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 19, 19, 64)        0         
_________________________________________________________________
flatten (Flatten)            (None, 23104)             0         
_________________________________________________________________
dense (Dense)                (None, 128)               2957440   
_________________________________________________________________
dense_1 (Dense)              (None, 1)                 129       
=================================================================
Total params: 2,976,385
Trainable params: 2,976,385
Non-trainable params: 0
_________________________________________________________________

Metrices

Confusion matrix

Loss

Accuracy