CNNs are a type of neural network designed for **image processing and pattern recognition**. They are widely used in tasks like **image classification, object detection, and medical imaging**.
### Key Layers in CNN
![[CNN.jpg]]
1️.**Convolution Layer**
- Extracts **features** (edges, textures) using **filters (kernels)**.
- Each filter slides over the input and applies a **dot product** operation.
2️.**Pooling Layer**
- Reduces **spatial dimensions** to improve efficiency.
- Common types:
- **Max Pooling** – Takes the largest value in a region.
- **Average Pooling** – Takes the average of values.
3️. **Fully Connected Layer**
- Flattens the feature map and connects to the final classification layer.
- Uses [[Activation Function]] (e.g., **Softmax** for multi-class classification).
### How CNNs Work ?
1. The **convolution layer** extracts features from the input image.
2. The **pooling layer** downsamples the image to reduce computation.
3. Multiple layers improve feature extraction.
4. The **fully connected layer** makes the final prediction.
### Why CNNs ?
- Learn **spatial hierarchies** (local and global patterns).
- Reduce parameters compared to fully connected networks.
- Work well with **images and videos**.