Deep learning Projects

Throughout my six years of experience as an artificial vision engineer, I have worked on multiple deep learning projects across various industries. Understanding how to apply this technology effectively is crucial, as it can be utilized in virtually any sector.

As a professional specializing in industrial and production environments, most of my projects have focused on:

Detecting defects in products on production lines.

Classify products on production lines.

Ensuring quality control to maintain high manufacturing standards.

Deep learning has proven to be a powerful tool in industrial applications, enabling more accurate, efficient, and adaptive inspection systems. My expertise in this field allows me to implement cutting-edge AI solutions that optimize production processes and enhance product quality.

Deep learning specialist

Software Deep learning in the industry

Vision Pro Deep learning

ViDi

Neuron

In-Sight Vision Suite

Throughout my experience with computer vision software, I have utilized various deep learning methodologies to develop and optimize applications. These techniques are essential for improving accuracy and solving complex industrial challenges.

The key methodologies I have worked with include:

OCV (Optical Character Verification) – Used for reading characters on products where conventional vision systems fail. (Cognex READ tool)
Classify – Applied for categorizing different types of products based on learned features.
Analyze – Allows manual defect annotation, enabling the system to recognize similar defects in future inspections.
Locate – Focuses on pattern detection to accurately find objects in an image.
Supervised Learning – Trained with labeled datasets (OK/KO) to improve defect detection.
Unsupervised Learning – Used when labeled data is limited, identifying anomalies based on deviations from normal patterns.

In addition to selecting the appropriate methodology, I ensure optimal dataset preparation, tuning key parameters, and applying evaluation metrics such as:

ROC curves to assess model performance.
Confusion matrices to analyze classification accuracy.

By combining deep learning techniques with rigorous data analysis and evaluation, I develop high-efficiency applications that enhance product quality and defect detection in industrial environments.

Passion for Deep learning

In my free time, I completed a Deep Learning specialization course, where I gained advanced knowledge in developing AI models to solve real-world problems across various industrial sectors.

Key Projects

During the course, I worked on six major challenges, using real-world datasets to develop innovative solutions:

Artificial Neural Networks (ANNs) for customer churn prediction.
Convolutional Neural Networks (CNNs) for image recognition.
Recurrent Neural Networks (RNNs) for stock price forecasting.
Self-Organizing Maps (SOMs) for fraud detection.
Boltzmann Machines for implementing recommendation systems.
Stacked AutoEncoders, applied to the Netflix million-dollar challenge.

Hands-On Approach

The course was highly practical and focused on coding from scratch, allowing me to develop my own implementations using Python and the most relevant Deep Learning frameworks:

TensorFlow and PyTorch, learning to compare their advantages and use cases.
Keras, for quickly and efficiently implementing complex models.
Theano, as an alternative for high-performance computations.
Scikit-learn, used for model evaluation through k-fold cross-validation, hyperparameter tuning, and data preprocessing.
Additional tools such as NumPy, Matplotlib, and Pandas for data manipulation and visualization.

Real-World Business Applications

Beyond theoretical and technical foundations, the course included practical applications in real business scenarios, including:

Banking & Finance: Customer churn prediction and fraud detection in credit card applications.
Healthcare: Medical image analysis for anomaly detection.
Retail & Entertainment: Building AI-powered recommendation systems.
AI emotions

This experience has strengthened my ability to apply Deep Learning techniques in industrial environments, optimizing processes and improving data-driven decision-making.

There are more information in my github: github.com/Jhon2704