user@devops:~$ cat README.md
CNN CIFAR-10
# Description
Image classification project with convolutional neural networks (CNN) on the CIFAR-10 dataset, containing 60,000 images in 10 categories (airplane, automobile, bird, cat, deer, dog, frog, horse, ship, truck). Implements a sequential CNN architecture with Conv2D, MaxPooling2D, Flatten and Dense layers, using ReLU and Softmax activation. Includes pixel normalization (0-1 scaling), data augmentation with ImageDataGenerator (rotation, shift, horizontal flip), Adam optimization and learning rate reduction. Features training metrics visualization (accuracy and loss) and confusion matrix for per-category evaluation.
# Key features
$ Image classification in 10 categories with from-scratch CNN
$ Architecture: Conv2D + MaxPooling2D + Dropout + Dense (Softmax)
$ Pixel normalization and data augmentation pipeline
$ Adam optimization with learning rate reduction on plateau
$ Metrics visualization: accuracy, loss and confusion matrix
$ Per-category evaluation with classification report
# Gallery
# Technologies used
