keras-io

Multiclass semantic segmentation using DeepLabV3+ This repo contains the model and the notebook to this Keras example on Multiclass semantic segmentation using DeepLabV3+. The model is trained for demonstrative purposes and does not guarantee the best results in production. For better results, follow & optimize the Keras example as per your need. Background Information Semantic segmentation, with the goal to assign semantic labels to every pixel in an image, is an essential computer vision task. In this example, we implement the DeepLabV3+ model for multi-class semantic segmentation, a fully-convolutional architecture that performs well on semantic segmentation benchmarks. Training Data The model is trained on a subset (10,000 images) of Crowd Instance-level Human Parsing Dataset. The Crowd Instance-level Human Parsing (CIHP) dataset has 38,280 diverse human images. Each image in CIHP is labeled with pixel-wise annotations for 20 categories, as well as instance-level identification. This dataset can be used for the "human part segmentation" task. Model The model uses ResNet50 pretrained on ImageNet as the backbone model. References: 1. Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation 2. Rethinking Atrous Convolution for Semantic Image Segmentation 3. DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs

Keras Implementation of OCR model for reading captcha 🤖🦹🏻 This repo contains the model and the notebook to this Keras example on OCR model for reading captcha. Background Information This example demonstrates a simple OCR model built with the Functional API. Apart from combining CNN and RNN, it also illustrates how you can instantiate a new layer and use it as an "Endpoint layer" for implementing CTC loss. This model uses subclassing, learn more about subclassing from this guide.

This repo contains the model and the notebook to this Keras example on Timeseries anomaly detection using an Autoencoder. This script demonstrates how you can use a reconstruction convolutional autoencoder model to detect anomalies in timeseries data. We will use the Numenta Anomaly Benchmark(NAB) dataset. It provides artifical timeseries data containing labeled anomalous periods of behavior. Data are ordered, timestamped, single-valued metrics. The following hyperparameters were used during training: - optimizer: {'name': 'Adam', 'learningrate': 0.001, 'decay': 0.0, 'beta1': 0.9, 'beta2': 0.999, 'epsilon': 1e-07, 'amsgrad': False} - trainingprecision: float32 | Epochs | Train Loss | Validation Loss | |--- |--- |--- | | 1| 0.011| 0.014| | 2| 0.011| 0.015| | 3| 0.01| 0.012| | 4| 0.01| 0.013| | 5| 0.01| 0.012| | 6| 0.009| 0.014| | 7| 0.009| 0.013| | 8| 0.009| 0.012| | 9| 0.009| 0.012| | 10| 0.009| 0.011| | 11| 0.008| 0.01| | 12| 0.008| 0.011| | 13| 0.008| 0.009| | 14| 0.008| 0.011| | 15| 0.008| 0.009| | 16| 0.008| 0.009| | 17| 0.008| 0.009| | 18| 0.007| 0.01| | 19| 0.007| 0.009| | 20| 0.007| 0.008| | 21| 0.007| 0.009| | 22| 0.007| 0.008| | 23| 0.007| 0.008| | 24| 0.007| 0.007| | 25| 0.007| 0.008| | 26| 0.006| 0.009| | 27| 0.006| 0.008| | 28| 0.006| 0.009| | 29| 0.006| 0.008| ## Model Plot

Tensorflow Keras Implementation of Multimodal entailment. This repo contains the models Multimodal Entailment. In this example, we will build and train a model for predicting multimodal entailment. We will be using the multimodal entailment dataset recently introduced by Google Research. On social media platforms, to audit and moderate content we may want to find answers to the following questions in near real-time: Does a given piece of information contradict the other? Does a given piece of information imply the other? In NLP, this task is called analyzing textual entailment. However, that's only when the information comes from text content. In practice, it's often the case the information available comes not just from text content, but from a multimodal combination of text, images, audio, video, etc. Multimodal entailment is simply the extension of textual entailment to a variety of new input modalities.