GitPlanet

Cheap and reliable Node.js hosting starts at $3/month, and $1/month static HTML hosting

Created with love in Canada, visit hostnodejs.com today

Feel like to post an Ad? Learn Details
All Projects → wangheda → Youtube 8m

wangheda / Youtube 8m

Licence: apache-2.0
The 2nd place Solution to the Youtube-8M Video Understanding Challenge by Team Monkeytyping (based on tensorflow)

Programming Languages

python
139335 projects - #7 most used programming language

Labels

deep-learningtensorflowcomputer-visionensemble-learningvideo-understanding

Projects that are alternatives of or similar to Youtube 8m

Predicting real estate prices using scikit Learn
Predicting Amsterdam house / real estate prices using Ordinary Least Squares-, XGBoost-, KNN-, Lasso-, Ridge-, Polynomial-, Random Forest-, and Neural Network MLP Regression (via scikit-learn)
Stars: ✭ 78 (-54.39%)
Mutual labels:  ensemble-learning
Dat8
General Assembly's 2015 Data Science course in Washington, DC
Stars: ✭ 1,516 (+786.55%)
Mutual labels:  ensemble-learning
Awesome Decision Tree Papers
A collection of research papers on decision, classification and regression trees with implementations.
Stars: ✭ 1,908 (+1015.79%)
Mutual labels:  ensemble-learning
Temporal Shift Module
[ICCV 2019] TSM: Temporal Shift Module for Efficient Video Understanding
Stars: ✭ 1,282 (+649.71%)
Mutual labels:  video-understanding
Handful Of Trials Pytorch
Unofficial Pytorch code for "Deep Reinforcement Learning in a Handful of Trials using Probabilistic Dynamics Models"
Stars: ✭ 112 (-34.5%)
Mutual labels:  ensemble-learning
I3d finetune
TensorFlow code for finetuning I3D model on UCF101.
Stars: ✭ 128 (-25.15%)
Mutual labels:  video-understanding
Temporally Language Grounding
A Pytorch implemention for some state-of-the-art models for" Temporally Language Grounding in Untrimmed Videos"
Stars: ✭ 73 (-57.31%)
Mutual labels:  video-understanding
Fox
Federated Knowledge Extraction Framework
Stars: ✭ 155 (-9.36%)
Mutual labels:  ensemble-learning
Movienet Tools
Tools for movie and video research
Stars: ✭ 113 (-33.92%)
Mutual labels:  video-understanding
Video2tfrecord
Easily convert RGB video data (e.g. .avi) to the TensorFlow tfrecords file format for training e.g. a NN in TensorFlow. This implementation allows to limit the number of frames per video to be stored in the tfrecords.
Stars: ✭ 137 (-19.88%)
Mutual labels:  video-understanding
Temporal Segment Networks
Code & Models for Temporal Segment Networks (TSN) in ECCV 2016
Stars: ✭ 1,287 (+652.63%)
Mutual labels:  video-understanding
Mlr3pipelines
Dataflow Programming for Machine Learning in R
Stars: ✭ 96 (-43.86%)
Mutual labels:  ensemble-learning
Mmaction
An open-source toolbox for action understanding based on PyTorch
Stars: ✭ 1,711 (+900.58%)
Mutual labels:  video-understanding
Xcessiv
A web-based application for quick, scalable, and automated hyperparameter tuning and stacked ensembling in Python.
Stars: ✭ 1,255 (+633.92%)
Mutual labels:  ensemble-learning
Awesome Activity Prediction
Paper list of activity prediction and related area
Stars: ✭ 147 (-14.04%)
Mutual labels:  video-understanding
Gcforest
This is the official implementation for the paper 'Deep forest: Towards an alternative to deep neural networks'
Stars: ✭ 1,214 (+609.94%)
Mutual labels:  ensemble-learning
Deepsuperlearner
DeepSuperLearner - Python implementation of the deep ensemble algorithm
Stars: ✭ 125 (-26.9%)
Mutual labels:  ensemble-learning
Object level visual reasoning
Pytorch Implementation of "Object level Visual Reasoning in Videos", F. Baradel, N. Neverova, C. Wolf, J. Mille, G. Mori , ECCV 2018
Stars: ✭ 163 (-4.68%)
Mutual labels:  video-understanding
Ensemble Pytorch
A unified ensemble framework for Pytorch to improve the performance and robustness of your deep learning model
Stars: ✭ 153 (-10.53%)
Mutual labels:  ensemble-learning
Multiverse
Dataset, code and model for the CVPR'20 paper "The Garden of Forking Paths: Towards Multi-Future Trajectory Prediction". And for the ECCV'20 SimAug paper.
Stars: ✭ 131 (-23.39%)
Mutual labels:  video-understanding
View All Similar Projects ➔

The Monkeytyping Solution to the Youtube-8M Video Understanding Challenge

This is the solution repository of the 2nd place team monkeytyping, licensed under the Apache License 2.0.

Dependencies

Python 2.7
Tensorflow 1.0
Numpy 1.12
GNU Bash

Resources

For an understanding of our system, read the report of our solution:

https://arxiv.org/abs/1706.05150

Our source code:

https://github.com/wangheda/youtube-8m

Useful scripts

Training scripts (training a model may take 3-5 days) are in

youtube-8m-wangheda/training_scripts
youtube-8m-zhangteng/train_scripts

Eval scripts for selecting best performing checkpoints

youtube-8m-wangheda/eval_scripts
youtube-8m-zhangteng/eval_scripts

Infer scripts for generating intermediate files used by ensemble scripts

youtube-8m-wangheda/infer_scripts
youtube-8m-zhangteng/infer_scripts

Ensemble scripts

youtube-8m-ensemble/ensemble_scripts

Paths of models and data

There are some conventions that we use in our code:

models are saved in

./model

train1 data is saved in

/Youtube-8M/data/frame/train
/Youtube-8M/data/video/train

validate1 data is saved in

/Youtube-8M/data/frame/validate
/Youtube-8M/data/video/validate

test data is saved in

/Youtube-8M/data/frame/test
/Youtube-8M/data/video/test

train2 data is saved in

/Youtube-8M/data/frame/ensemble_train
/Youtube-8M/data/video/ensemble_train

validate2 data is saved in

/Youtube-8M/data/frame/ensemble_validate
/Youtube-8M/data/video/ensemble_validate

intermediate results are stored in

/Youtube-8M/model_predictions/ensemble_train/[method]
/Youtube-8M/model_predictions/ensemble_validate/[method]
/Youtube-8M/model_predictions/test/[method]

How to generate a solution

Single model

  1. Train a single model
  2. evaluate the checkpoints to get the best one
  3. infer the checkpoint to get intermediate result.

Ensemble model

  1. Write a configuration file
  2. train a stacking model
  3. evaluate the stacking model and pick the best checkpoint
  4. infer the checkpoint to get a submission file

Note

Some of the single models are developed by Heda and some by Teng, so they are distributed in two folders.

Bagging models are in youtube-8m-wangheda/bagging_scripts.

Boosting and distillation models are in youtube-8m-wangheda/bagging_scripts.

Cascade models are in youtube-8m-wangheda/cascade_scripts.

Stacking models are in youtube-8m-ensemble/ensemble_scripts.

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].