preview

Applications Of Cnn In Computer Vision Essay

Decent Essays
Open Document

Applications of CNN in Computer Vision
Applications of CNN in Computer Vision Computer vision is a very broad research area which covers a wide variety of approaches not only to process images but also to understand their contents. It is an active research field for convolutional neural network applications. The most popular of these applications include, classification, segmentation, detection and scene understanding. Most CNN architectures have been used for computer vision problems including, supervised or unsupervised face/object classification (e.g., to identify an object or a person in a given image or to output the class label of that object), detection (e.g., annotate an image with bounding boxes around each object), segmentation …show more content…

While this CNN model has significantly reduced the error rate for image classification, we shall discuss below more recent and advanced deep CNN architectures, which have achieved a very high classification performance.
1.1. PointNet
PointNet [3] is a new type of neural network which directly consumes point clouds and well respects the permutation invariance of the points in the input image.
PointNet, shown in Fig. 1, provides a unified architecture for applications ranging from object classification, and part segmentation, to scene semantic parsing from pointclouds. It directly takes point clouds as input and outputs either class labels for the entire input, or per point segment labels for each point of the input. PointNet has three main modules, which we briefly discuss below.
2
1.1.1. Symmetric Function (Max Pooling Layer - Module A):
The first key module is the max pooling layer, used as a symmetric function to aggregate information from all the points, and to make a model invariant to input permutations. To achieve this, a general function is defined on a point set by applying a symmetric function on the transformed elements in the set: f(x1, . . . , xn) ≈ g(h(x1), . . . , h(xn)), (1)
In the above equation, f : 2RN → R, h : RN → RK and g : RK × · · · × RK
→R is a symmetric function. h is approximated by a multi-layer perceptron (mlp) network and g by a composition of a single variable function and a max pooling function. Through a collection of

Get Access
Related
Get Access