Alexnet is a image classification model where the input is an image of one of 1000 different classes (e.g. cats, dogs etc.) and the output is a vector of 1000 numbers. The ith element of the output vector is the probability that the input image belongs to the ith class; therefore, the sum of all elements of the output vector is 1.The input to AlexNet is an RGB image of size 224x224 px.
EfficientNet-Lite4 is for image classification. It achieves high accuracy and can operate on a mobile CPU (in addition to GPU, and TPU) where computational resources are limited. The input is an RBG image with the size of (224 x 224 px) x 3 channels. Output of the model is an array score with the length of 1000.
This model is for recognizing emotion in faces. Facial expression recognition is done with crowd-sourced label distribution. The model requires image inputs with dimensions of 64 x 64 px. The model outputs an array of 8 scores corresponding to 8 emotion classes: emotion_table = ['neutral':0, 'happiness':1, 'surprise':2, 'sadness':3, 'anger':4, 'disgust':5, 'fear':6, 'contempt':7]
ResNet50 is a convolutional neural network with 50 layers used for image classification. Deep neural networks are difficult to train; residual blocks in this model facilitate the training of deep networks; make the optimization process easier and increase its accuracy. Data input is an image (224 x 224 px) x 3 channels and the output is an array of length 1000. This network was trained on Imagenet.
This script is used for classification of the Iris Dataset using RandomForestClassifier from sklearn. The data set contains 3 classes of 50 instances each, where each class refers to a type of iris plant. The first class is linearly separable from the others, the latter two are NOT linearly separable from each other.
Neural style transfer is an optimization technique used to alter a content image in the style of a reference image, such as an artwork by a famous painter. The resulting and blend produces an output image where the content image appears in the styleof the style-reference image. This is achieved by optimizing the output image to match the content statistics of the content image and the style-reference image.
Super resolution is the process of upscaling and or improving the details within an image. Often a low resolution image is taken as an input and the subsequent output image is upscaled to a higher resolution. The details in the high resolution output are a prediction added in where the details are unknown.
This convolutional neural network is ideal for image segmentation tasks in biomedical applications. Attributes of its architecture require fewer training images to obtain precise segmentations. Input images of 512 x 512 px are typical.
Object detection is a computer vision task that predicts the presence of one or more objects, along with their classes and bounding boxes in video or image inputs. YOLO (You Only Look Once) is a state-of-the-art Object Detector that conducts object detection in real-time with high accuracy. YOLOv3 is a real-time, single-stage object detection model that can detects 80 different classes.
Get started with the free Zetane Viewer (no email required).