9 – 09 Detection Without Proposals Summary V1

You’ve really learned a lot about the evolution of CNN architectures in this lesson, especially about how to detect multiple objects in an image using region proposals. As we mentioned at the start, simple image classification problems are really the most basic application for CNNs. And if we’re aiming to reach levels of human understanding, … Read more

8 – 08 Faster RCNN V1 RENDER V2

To speed up the time it takes to run a test image through a network and detect all the objects in it, we want to decrease the time it takes to form region proposals. For this we have the faster R-CNN architecture. Faster R-CNN learns to come up with its own region proposals. It takes … Read more

7 – 07 Fast RCNN V1 RENDER V2

The next advancement in region-based CNNs came with the Fast R-CNN architecture. Instead of processing each region of interest individually through a classification CNN, this architecture runs the entire image through a classification CNN only once. The image goes through a series of convolutional and pooling layers and at the end of these layers, we … Read more

6 – 06 RCNN V1 RENDER V2

To localize and classify multiple objects in an image, we want to be able to identify a limited set of cropped regions for a CNN to look at. In the ideal case, we would generate three perfectly cropped regions for three different objects in an image. To approach this goal and generate a good limited … Read more

5 – 05 Region Proposals V1 RENDER V2

Now you’ve seen how to locate one object in an image by generating a bounding box around that object. But what if there are multiple objects in an image? How can you train a network to detect all of them? Well, let’s think about the case where we just have two objects in an image. … Read more

4 – 04 Bounding Boxes And Regression V1 RENDER V3

When we train a CNN to classify a set of images, we train it by comparing the output predicted class with the true class label and seeing if they match. We typically use Cross-entropy to measure the error between these classes because Cross-entropy loss decreases as the predicted class which has some uncertainty associated with … Read more

3 – 03 Classification And Localization RENDER V3

In classification tasks we’ve seen. We give an image to a CNN and it outputs a label for that entire image. But sometimes you want a little more information. Say where the object actually is located in the image and this is called localization. For example, say you have an image of basketball players and … Read more

2 – 02 More Than Classification RENDER V2

In this lesson we’ll talk about a few different types of CNN architectures that can do more than single image classification. First we’ll talk about classifying one object in an image and localizing it, which means finding its location in the image. This is typically done by placing a bounding box around the object. Finding … Read more

1 – 01 CNNs And Scene Understanding RENDER Full V2

So far, you’ve seen a variety of image processing techniques that play a foundational role in pattern recognition tasks, such as an image classification. You’ve seen how convolutional neural networks follow a series of steps to classify an image. Just to recap, a CNN first takes in an input image then puts that image through … Read more