Video Captioning

Generate caption for the given video clip

Branch : VideoCaption (1a2124d), VideoCaption_catt (647e73b4)

Model

Model generates natural sentence word by word

Audio SubModel	Video SubModel	Sentence Generation SubModel

Context extraction for Temporal Attention Model, at i^th word generation

Results - f5c22f7

Test videos with good results

two men are talking about a cooking show	a woman is cooking	a dog is running around a field
a woman is talking about a makeup face	a man is driving a car down the road	a man is cooking in a kitchen
a man is playing a video game	two men are playing table tennis in a stadium	a man is talking about a computer program

Test videos with poor results

a person is playing with a toy	a man is walking on the field	a man is standing in a gym

Try it out!!!

• Please feel free to raise PR with necessary suggestions.

• Clone the repository`

  • git clone https://github.com/scopeInfinity/Video2Description.git

• Install docker and docker-compose

  • Current config has docker-compose file format '3.2'.
    • https://github.com/docker/compose/releases

  • sudo apt-get install docker.io
    sudo curl -L "https://github.com/docker/compose/releases/download/1.25.4/docker-compose-$(uname -s)-$(uname -m)" -o /usr/local/bin/docker-compose
    sudo chmod +x /usr/local/bin/docker-compose

  • docs
    • https://docs.docker.com/install/linux/docker-ce/ubuntu/
    • https://docs.docker.com/compose/install/

• Pull the prebuild images and run the container

$ docker-compose pull
$ docker-compose up

• Browse to http://localhost:8080/
  • backend might take few minutes to reach a stable stage.

Execution without Docker

• We can go always go through backend.Dockerfile and frontend.Dockerfile to understand better.
• Update src/config.json as per the requirement and use those path during upcoming steps.
  • To know more about any field, just search for the reference in the codebase.
• Install miniconda
  • https://docs.conda.io/projects/conda/en/latest/user-guide/install/index.html
• Get glove.6B.300d.txt from https://nlp.stanford.edu/projects/glove/
• Install ffmpeg
  • Configure, build and install ffmpeg from source with shared libraries

$ git clone 'https://github.com/FFmpeg/FFmpeg.git'
$ cd FFmpeg
$ ./configure --enable-shared  # Use --prefix if need to install in custom directory
$ make
# make install

• If required, use https://github.com/tylin/coco-caption/ for scoring the model.
• Then create conda environment using environment.yml
  • $ conda env create -f environment.yml
• And activate the environment

$ conda activate .

• Turn up the backend
  • src$ python -m backend.parser server --start --model /path/to/model
• Turn up the web frontend
  • src$ python -m frontend.app

Info

Data Directory and Working Directory can be same as the project root directory.

Data Directory

Working Directory

Download Dataset

• Execute python videohandler.py from VideoDataset Directory

Execution

It currently supports train, predict and server mode. Please use the following command for better explanation.

src$ python -m backend.parse -h

Training Methods

• Try Iterative Learning
• Try Random Learning

Evaluation

Prerequisite

cd /path/to/eval_dir/
git clone 'https://github.com/tylin/coco-caption.git' cococaption
ln /path/to/working_dir/cocoeval.py cococaption/

Evaluate

# One can do changes in parser.py for numbers of test examples to be considered in evaluation
python parser.py predict save_all_test
python /path/to/eval_dir/cocoeval.py <results file>.txt

Sample Evaluation while training

Check Specifications section for model comparision.

Temporal attention Model for is on VideoCaption_catt branch.

Pre-trained Models : https://drive.google.com/open?id=1gexBRQfrjfcs7N5UI5NtlLiIR_xa69tK

Web Server

• Start the server (S) for to compute predictions (Within conda environment)

python parser.py server -s -m <path/to/correct/model>

• Check config.json for configurations.
• Execute python app.py from webserver (No need for conda environment)
  • Make sure, your the process is can new files inside $UPLOAD_FOLDER
• Open http://webserver:5000/ to open Web Server for testing (under default configuration)

Specifications

Commit: 3ccf5d5

• ResNet over LSTM for feature extraction
• Word by Word generation based on last prediction for Sentence Generation using LSTM
• Random Dataset Learning of training data
• Vocab Size 9448
• Glove of 300 Dimension

Commit: bd072ac

• ResNet over BiDirection GRU for feature extraction
• Sequential Learning of training data
• Batch Normalization + Few more tweaks in Model
• Bleu, CIDEr, Rouge, Meteor score generation for validation
• Multiprocessing keras

Commit: f5c22f7

• Audio with BiDirection GRU

Commit: 1a2124d

• Audio with BiDirection LSTM

Commit: 647e73b

• Audio with BiDirection GRU using temporal attention for context

Image Captioning

Generate caption for the given images

Branch : onehot_gen

Commit : 898f15778d40b67f333df0a0e744a4af0b04b16c

Trained Model : https://drive.google.com/open?id=1qzMCAbh_tW3SjMMVSPS4Ikt6hDnGfhEN

Categorical Crossentropy Loss : 0.58