From the course: Deep Learning with Python: Convolutional Neural Networks
Join today to access over 24,900 courses taught by industry experts.
Why are CNNs so effective for computer vision? - Python Tutorial
From the course: Deep Learning with Python: Convolutional Neural Networks
Why are CNNs so effective for computer vision?
- [Instructor] Now that we know how Convolutional Neural Networks work, it is important to understand why they are preferred over traditional deep neural networks for image data and Computer Vision. Traditional deep learning models are fully connected, meaning every neuron in one layer connects to every neuron in the next. For image data, this becomes impractical very quickly. Consider a single 256 by 256 pixel RGB image. This image has 196,608 input values, 256 pixels times 256 pixels times three channels. Feeding this image, as is, into a fully connected layer means that a single neuron in the first hidden layer would need 196,608 weights, and that is only for one neuron. When you scale this to an entire layer of neurons, the number of parameters increases geometrically, and if the image were larger, for example, 1024 pixels by 1024 pixels, we would need 9,437,184 weights in the first layer. Such a model would not only be prone to overfitting, it would also be computationally…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.