All Projects → oswaldoludwig → Human Action Recognition With Keras

oswaldoludwig / Human Action Recognition With Keras

Licence: gpl-3.0
Keras implementation of Human Action Recognition for the data set State Farm Distracted Driver Detection (Kaggle)

Programming Languages

python
139335 projects - #7 most used programming language

Projects that are alternatives of or similar to Human Action Recognition With Keras

Artificial Intelligence Deep Learning Machine Learning Tutorials
A comprehensive list of Deep Learning / Artificial Intelligence and Machine Learning tutorials - rapidly expanding into areas of AI/Deep Learning / Machine Vision / NLP and industry specific areas such as Climate / Energy, Automotives, Retail, Pharma, Medicine, Healthcare, Policy, Ethics and more.
Stars: ✭ 2,966 (+1801.28%)
Mutual labels:  kaggle, convolutional-neural-networks
Data Science Bowl 2018
End-to-end one-class instance segmentation based on U-Net architecture for Data Science Bowl 2018 in Kaggle
Stars: ✭ 56 (-64.1%)
Mutual labels:  kaggle, convolutional-neural-networks
Facial Expression Recognition
Classify each facial image into one of the seven facial emotion categories considered using CNN based on https://www.kaggle.com/c/challenges-in-representation-learning-facial-expression-recognition-challenge
Stars: ✭ 82 (-47.44%)
Mutual labels:  kaggle, convolutional-neural-networks
Binary Face Alignment
Real time face alignment
Stars: ✭ 145 (-7.05%)
Mutual labels:  convolutional-neural-networks
Fast Neural Style Pytorch
Fast Neural Style Transfer implementation in PyTorch 🎨 🎨 🎨
Stars: ✭ 146 (-6.41%)
Mutual labels:  convolutional-neural-networks
Mimick
Code for Mimicking Word Embeddings using Subword RNNs (EMNLP 2017)
Stars: ✭ 152 (-2.56%)
Mutual labels:  convolutional-neural-networks
Sparse Winograd Cnn
Efficient Sparse-Winograd Convolutional Neural Networks (ICLR 2018)
Stars: ✭ 156 (+0%)
Mutual labels:  convolutional-neural-networks
Livianet
This repository contains the code of LiviaNET, a 3D fully convolutional neural network that was employed in our work: "3D fully convolutional networks for subcortical segmentation in MRI: A large-scale study"
Stars: ✭ 143 (-8.33%)
Mutual labels:  convolutional-neural-networks
Open Solution Toxic Comments
Open solution to the Toxic Comment Classification Challenge
Stars: ✭ 154 (-1.28%)
Mutual labels:  kaggle
Regnet
Nonrigid image registration using multi-scale 3D convolutional neural networks
Stars: ✭ 151 (-3.21%)
Mutual labels:  convolutional-neural-networks
Tfvos
Semi-Supervised Video Object Segmentation (VOS) with Tensorflow. Includes implementation of *MaskRNN: Instance Level Video Object Segmentation (NIPS 2017)* as part of the NIPS Paper Implementation Challenge.
Stars: ✭ 151 (-3.21%)
Mutual labels:  convolutional-neural-networks
Self Driving Golf Cart
Be Driven 🚘
Stars: ✭ 147 (-5.77%)
Mutual labels:  convolutional-neural-networks
Compactcnncascade
A binary library for very fast face detection using compact CNNs.
Stars: ✭ 152 (-2.56%)
Mutual labels:  convolutional-neural-networks
Concrete Crack Detection
Crack detection for an autonomous UAV
Stars: ✭ 146 (-6.41%)
Mutual labels:  convolutional-neural-networks
Deep Viz Keras
Implementations of some popular Saliency Maps in Keras
Stars: ✭ 154 (-1.28%)
Mutual labels:  convolutional-neural-networks
Voicemap
Identifying people from small audio fragments
Stars: ✭ 144 (-7.69%)
Mutual labels:  convolutional-neural-networks
Pytorch Question Answering
Important paper implementations for Question Answering using PyTorch
Stars: ✭ 154 (-1.28%)
Mutual labels:  convolutional-neural-networks
Pins
Pin, Discover and Share Resources
Stars: ✭ 149 (-4.49%)
Mutual labels:  kaggle
Models Comparison.pytorch
Code for the paper Benchmark Analysis of Representative Deep Neural Network Architectures
Stars: ✭ 148 (-5.13%)
Mutual labels:  convolutional-neural-networks
Deepvideodeblurring
S. Su, M. Delbracio, J. Wang, G. Sapiro, W. Heidrich, O. Wang. Deep Video Deblurring. CVPR 2017, Spotlight
Stars: ✭ 151 (-3.21%)
Mutual labels:  convolutional-neural-networks

Human-Action-Recognition-with-Keras

Keras implementation of Human Action Recognition for the data set State Farm Distracted Driver Detection (Kaggle).

DESCRIPTION:

This model uses 3 dense layers on the top of the convolutional layers of a pre-trained ConvNet (VGG-16) to classify driver actions into 10 classes. The dense layers are subject to an aggressive regularization by Eigenvalue Decay, the last 7 convolutional layers of the VGG-16 are fine tuned and data augmentation was applied on the training data (shear, zoom and rotation).

This model was ranked in the top 29% in the Kaggle private leaderboard after 80 training epochs. I believe the performance of this model could be much better if the model was trained during a larger number of epochs (since I adopted an aggressive regularization).

This script reuses pieces of code from the post "Building powerful image classification models using very little data", from blog.keras.io, and from https://www.kaggle.com/tnhabc/state-farm-distracted-driver-detection/keras-sample

INSTALLATION AND USE:

  1. Download the training data at https://www.kaggle.com/c/state-farm-distracted-driver-detection/data (keeping the name of the folders as "train" and "test");

  2. Download the file "vgg16_weights.h5" containing the pre-trained weights of the VGG-16 at https://gist.github.com/baraldilorenzo/07d7802847aaad0a35d3

  3. Set the path for the file "vgg16_weights.h5" in line 32 of the code, variable "weights_path", and the number of epochs in line 57, I suggest more than 80 epoch, if you can;

  4. This model uses the Eigenvalue Decay regularizer, if Keras is already installed, check if you have this regularizer, if you don't have it, update Keras to have Eigenvalue Decay: sudo pip install git+git://github.com/fchollet/keras.git --upgrade

  5. Run “HumanActionRecognition.py” to train the deep model and create the submission file with the estimated classes for the test data. To run in GPU you can call the code like this: THEANO_FLAGS=mode=FAST_RUN,device=gpu,floatX=float32,exception_verbosity=high python HumanActionRecognition.py

Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].