Several tutorials are provided in order to dive into DeepDetect.
All tutorials are explicitely crafted so that they do not require data or models that are not publicly available. Thus anyone should be able to reproduce the reported results and applications.
List of tutorials:
Setup an image classifier: tag any image from the Web with state of the art Deep Learning
Setup an object detector: predict a bounding box around objects of interest and tag each of them
Train a service from a CSV dataset: very common generic use case, use a multilayer neural network, here to predict trees’ cover type
Train your own image classifier: use state of the art Deep Learning to build a service that recognizes images
Train from text: use a multilayer neural network to classify text files into several categories
Multi-GPU training: use multiple GPUs to scale up the training or large deep neural networks
Robust image model training with Data Augmentation: use random noise and distortions of images to make image classification and object detection models more robust, e.g. to user generated content
Tag images into ElasticSearch: build an image search engine using a deep convolutional neural network to classify images into categories and send them to ElasticSearch with no additional code involved