Use unprefixed input range selectors in CSS
Auto-DL helps you make Deep Learning models without writing a single line of code and giving as little input as possible.
Captionify.me is a Deep Learning based single page Angular application that predicts captions for a given input image. The image captioning model has been trained on MS-COCO 2014 training dataset. In this project the Merge encoder-decoder model has been practically implemented to generate caption. The TTS component of the web application synthesizes the speech based on the output of the model.
Deep learning model for single-cell inference of multi-omic profiles from a single input modality.
Diseases in the leaves of plants are very crucial issue in this day and due to which yield of high-quality crops gets devasted, and the longevity of the plant hampers. And also, it is very difficult to understand the current condition of the leaf with the naked eye. Which results in the reduction of yield of high-quality crops. To overcome this problem, we planned to use machine learning based approach to segment, to select every small part of the leaf and detect the disease, also to analyse the quality. The main vision of the paper is to detect all possible diseases of the leaves of the plant by applying Neural Network for classify the disease based on the colour changes in comparison to the analytical available data set, providing the best fit output. In this proposed work, Apple plant leaf dataset used which contain 1910 images of healthy and unhealthy leaves. These leaves are pre-processed first through some steps. In this pre-processing convert the RGB image with the help of BGR2GRAY function available on open cv library then GaussianBlur function is used to remove the additional noises and smoothing of the edges of the leaves which helped to detect the main object or leaf more accurately from the image in future. After smoothing threshold function is applied for removing the unnecessary background from the image especially thresh_torezo_inv from cv2 library used for this work it helped to convert the unnecessary background into a single colour and focus on the main object i.e. leaf. After thresholding Erode and Dilate functions are applied for more cleaner image. Then with the help of FindContours() function from cv2 library helped to find four extreme points of the desired object i.e. leaf and crop the image according to the extreme points. Then the images are resized into 240, 240px size, it is necessary all the images are in same size for the best result output. Then each image is transformed to an n-dimensional array and appended into a list. After this pre-processing the data set is splitted into 3 different parts, i.e. training, validation and testing. 1337 images are used for training i.e. training part contains 1337 images, 287 images are used for validation and 286 images are used for testing. After splitting the dataset, the main convolutional neural network model is constructed. In the convolutional neural network model one input layer used and two hidden layers used and one out put layer used. In the hidden layer 32 nodes each are used with normal weight initialization and ReLu function for Activation. And in the output layer 1 node is used and sigmoid function for activation. On the compilation of the model stochastic gradient descent used for optimizing with a learning rate 0.05 and loss being calculated using categorical_crossentropy. After compilation of the model it is fitted into the training dataset with batch size of 32 and 10 epochs.
A scratch-built neural network training and application suite. I implemented a customizable deep-neural network to train against randomized vector data to begin to experiment with video compression, wherein input seed vectors are smaller than the output vectors. I built a Genetic Algorithm as the primary training algorithm but also inluded experimental implementations of Particle Swarm Optimization and Ensemble Learning. The Genetic Algorithm is multi-threaded and I also attempted to harness GPU matrix evaluation using Aparapi but never completed it. The Genetic Algorithm manipulates the weights as an analog to DNA. At each round the DNA is evaluated by applying its properties to a network and running that network against the training data. Experiments can be run by changing which enum constants are used in Driver and by changing parameters in the training algorithms. Feel free to experiment and explore!
Build forms in React, without the tears
A trading bitcoin agent was created with deep reinforcement learning implementations. Various experiments were performed on the type of neural network, the type of reinforcement learning algorithm and the number of daily input values that initially required from agent to make the first decision.
A Flask and Deep Learning Project that recognizes the emotion/mood of the user via either a photo or voice or text given as input by the user. It also contains a chatbot!
A recent study has shown that around 75% of criminal investigations go unsolved. Only 1 out of 3 criminals gets arrested in America. According to recent statistics from Red Cross in 2018, the number of people who went missing worldwide are around 100,000. With so many people missing and so many cases going unsolved for not being able to fully identify the person’s face and facial characteristics. The Science of Deep learning offers a variety of solutions. Deep learning is a branch of machine learning which is completely based on artificial neural networks, as neural network is going to mimic the human brain so deep learning is also a kind of mimic of human brain. It works on datasets with huge amounts of data. Given a large dataset of input and output pairs, a deep learning algorithm will try to minimize the difference between its prediction and expected output. By doing this, it tries to learn the association/pattern between given inputs and outputs, this in turn allows a deep learning model to generalize to inputs that it hasn’t seen before. So, how do deep learning Algorithms learn? Deep Learn- 4 ing Algorithms use something called a neural network to find associations between a set of inputs and outputs. The “deep” part of deep learning refers to creating deep neural networks. This refers to a neural network with a large quantity of layers, with the addition of more weights and biases, the neural network improves its ability to approximate more complex functions. Which in return generates a more accurate output. So , This project aims to make an application that takes a sketch image as an input and generates a realistic face image as an output. Also Creating realistic human face images from sketches can be used for various applications including criminal investigation, character design, educational training, etc.