theshadow29 / zsgnet-pytorch Goto Github PK

Official implementation of ICCV19 oral paper Zero-Shot grounding of Objects from Natural Language Queries (https://arxiv.org/abs/1908.07129)

License: MIT License

Python 99.09% Shell 0.91%

grounding vision nlp objects

zsgnet-pytorch's Introduction

zsgnet-pytorch

This is the official repository for ICCV19 oral paper Zero-Shot Grounding of Objects from Natural Language Queries. It contains the code and the datasets to reproduce the numbers for our model ZSGNet in the paper.

The code has been refactored from the original implementation and now supports Distributed learning (see pytorch docs) for significantly faster training (around 4x speedup from pytorch Dataparallel)

The code is fairly easy to use and extendable for future work. Feel free to open an issue in case of queries.

Install

Requirements:

python>=3.6
pytorch>=1.1

To use the same environment you can use conda and the environment file conda_env_zsg.yml file provided. Please refer to Miniconda for details on installing conda.

MINICONDA_ROOT=[to your Miniconda/Anaconda root directory]
conda env create -f conda_env_zsg.yml --prefix $MINICONDA_ROOT/envs/zsg_pyt
conda activate zsg_pyt

Data Preparation

Look at DATA_README.md for quick start and DATA_PREP_README.md for obtaining annotations from the parent datasets.

Training

Basic usage is python code/main_dist.py "experiment_name" --arg1=val1 --arg2=val2 and the arg1, arg2 can be found in configs/cfg.yaml. This trains using the DataParallel mode.

For distributed learning use python -m torch.distributed.launch --nproc_per_node=$ngpus code/main_dist.py instead. This trains using the DistributedDataParallel mode. (Also see caveat in using distributed training below)

An example to train on ReferIt dataset (note you must have prepared referit dataset) would be:

python code/main_dist.py "referit_try" --ds_to_use='refclef' --bs=16 --nw=4

Similarly for distributed learning (need to set npgus as the number of gpus)

python -m torch.distributed.launch --nproc_per_node=$npgus code/main_dist.py "referit_try" --ds_to_use='refclef' --bs=16 --nw=4

Logging

Logs are stored inside tmp/ directory. When you run the code with $exp_name the following are stored:

txt_logs/$exp_name.txt: the config used and the training, validation losses after ever epoch.
models/$exp_name.pth: the model, optimizer, scheduler, accuracy, number of epochs and iterations completed are stored. Only the best model upto the current epoch is stored.
ext_logs/$exp_name.txt: this uses the logging module of python to store the logger.debug outputs printed. Mainly used for debugging.
tb_logs/$exp_name: this is still wip, right now just creates a directory and nothing more, ideally want to support the tensorboard logs.
predictions: the validation outputs of current best model.

Evaluation

There are two ways to evaluate.

For validation, it is already computed in the training loop. If you just want to evaluate on validation or testing on a model trained previously ($exp_name) you can do:

python code/main_dist.py $exp_name --ds_to_use='refclef' --resume=True --only_valid=True --only_test=True

or you can use a different experiment name as well and pass --resume_path argument like:

python code/main_dist.py $exp_name --ds_to_use='refclef' --resume=True --resume_path='./tmp/models/referit_try.pth'

After this, the logs would be available inside tmp/txt_logs/$exp_name.txt

If you have some other model, you can output the predictions in the following structure into a pickle file say predictions.pkl:

[
    {'id': annotation_id,
 	'pred_boxes': [x1,y1,x2,y2]},
    .
    .
    .
]

Then you can evaluate using code/eval_script.py using:

python code/eval_script.py predictions_file gt_file

For referit it would be

python code/eval_script.py ./tmp/predictions/$exp_name/val_preds_$exp_name.pkl ./data/referit/csv_dir/val.csv

Caveats in DistributedDataParallel

When training using DDP, there is no easy way to get all the validation outputs into one process (that works only for tensors). As a result one has to save the predictions of each separate process and then read again to combine them in the main process. Current implementation doesn't do this for simplicity, as a result the validation results obtained during training are slight different from the actual results.

To get the correct results, one can follow the steps in Evaluation as is (the point to note is NOT use torch.distributed.launch for evaluation). Thus, you would get correct results when using simply dataparallel.

Pre-trained Models

The pre-trained models are available in Google Drive

ToDo

Acknowledgements

We thank:

@yhenon for their repository on retina-net (https://github.com/yhenon/pytorch-retinanet).
@amdegroot for their repsository on ssd using vgg (https://github.com/amdegroot/ssd.pytorch)
fastai repository for helpful logging, anchor box generation and convolution functions.
maskrcnn-benchmark repository for many of the distributed utils and implementation of non-maxima suppression.

Citation

If you find the code or dataset useful, please cite us:

@InProceedings{Sadhu_2019_ICCV,
author = {Sadhu, Arka and Chen, Kan and Nevatia, Ram},
title = {Zero-Shot Grounding of Objects From Natural Language Queries},
booktitle = {The IEEE International Conference on Computer Vision (ICCV)},
month = {October},
year = {2019}
}

zsgnet-pytorch's People

Contributors

Stargazers

Watchers

Forkers

luogen1996 youngfly11 ruotianluo oujieww jinjing326 s-hisenberg suo-wei fanhhhhh mengqidyangge nimo1989 linhuixiao daniel-koudouna

zsgnet-pytorch's Issues

incompatible in install using miniconda

I try to use your yml file conda_env_zsg.yml but failed with following error. (and can you give less package in your yml file to avoid imcompatible)?

UnsatisfiableError: The following specifications were found to be incompatible with each other:

Package sqlite conflicts for:
ipywidgets==7.4.2=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
ipykernel==5.1.0=py37h24bf2e0_1002 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
prometheus_client==0.6.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pip==19.1=py37_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
setuptools==41.0.1=py37_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
murmurhash==1.0.0=py37hf484d3e_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
thinc==7.0.4=py37hc9558a2_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
plac==0.9.6=py_1 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
llvmlite==0.29.0=py37hfd453ef_1 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
wasabi==0.2.2=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pygments==2.4.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
wcwidth==0.1.7=py_1 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
bz2file==0.98=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
jsonschema==3.0.0a3=py37_1000 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
jedi==0.13.3=py37_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
wheel==0.33.2=py37_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
scikit-learn==0.21.2=py37h627018c_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
tornado==6.0.2=py37h516909a_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
srsly==0.0.7=py37he1b5a44_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
cryptography==2.6.1=py37h72c5cf5_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
terminado==0.8.2=py37_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
six==1.12.0=py37_1000 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
parso==0.4.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
gensim==3.7.3=py37he1b5a44_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
idna==2.8=py37_1000 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
mistune==0.8.4=py37h14c3975_1000 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
traitlets==4.3.2=py37_1000 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
libwebp==1.0.2=h576950b_1 -> libtiff[version='>=4.0.9,<5.0a0'] -> zstd[version='>=1.3.7,<1.3.8.0a0'] -> lz4 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='3.13.|3.20.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
nltk==3.2.5=py_0 -> python -> sqlite[version='3.20.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pyrsistent==0.15.1=py37h516909a_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pyjwt==1.7.1=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
python-dateutil==2.8.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pandocfilters==1.4.2=py_1 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
qtconsole==4.4.4=py_0 -> python -> sqlite[version='3.20.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pycparser==2.19=py37_1 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
jupyter_core==4.4.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
markupsafe==1.1.1=py37h14c3975_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
python==3.7.3=h5b0a415_0 -> sqlite[version='>=3.26.0,<4.0a0']
jupyter_console==6.0.0=py_0 -> python[version='>=3.5'] -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
asn1crypto==0.24.0=py37_1003 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
cffi==1.12.3=py37h8022711_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
jinja2==2.10.1=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
boto3==1.9.158=py_0 -> python -> sqlite[version='3.13.|3.20.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
ipython_genutils==0.2.0=py_1 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
urllib3==1.24.3=py37_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pysocks==1.7.0=py37_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pyzmq==18.0.1=py37hc4ba49a_1 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pytz==2019.1=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
olefile==0.46=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
python-crfsuite==0.9.6=py37h6bb024c_1000 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
sqlite==3.26.0=h67949de_1001
backcall==0.1.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
webencodings==0.5.1=py_1 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
attrs==19.1.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
cymem==2.0.2=py37hfd86e86_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
notebook==5.7.8=py37_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
matplotlib-base==3.1.0=py37hfd891ef_1 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pexpect==4.7.0=py37_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
defusedxml==0.5.0=py_1 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
torchvision==0.2.2=py_3 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
spacy==2.1.4=py37hc9558a2_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
opencv==4.1.1=py37hd64ca61_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
scipy==1.3.0=py37h921218d_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
decorator==4.4.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
nbformat==4.4.0=py_1 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pyqt==5.9.2=py37hcca6a23_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pillow==6.1.0=py37h6b7be26_1 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
s3transfer==0.2.0=py37_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
preshed==2.0.1=py37he6710b0_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
numpy==1.16.3=py37he5ce36f_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
oauthlib==3.0.1=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
ipython==7.5.0=py37h24bf2e0_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
smart_open==1.8.3=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
requests-oauthlib==1.2.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
chardet==3.0.4=py37_1003 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pytorch==1.1.0=py3.7_cuda9.0.176_cudnn7.5.1_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
sip==4.19.8=py37hf484d3e_1000 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
requests==2.22.0=py37_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
boto==2.49.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
jmespath==0.9.4=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
blinker==1.4=py_1 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
ptyprocess==0.6.0=py_1001 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
twython==3.7.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
send2trash==1.5.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pandas==0.24.2=py37hf484d3e_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
nbconvert==5.5.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
joblib==0.13.2=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
jupyter_client==5.2.4=py_3 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
docutils==0.14=py37_1001 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
prompt_toolkit==2.0.9=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pickleshare==0.7.5=py37_1000 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
bleach==3.1.0=py_0 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
pyopenssl==19.0.0=py37_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
numba==0.45.1=py37hb3f55d8_0 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
jupyter==1.0.0=py_2 -> python -> sqlite[version='3.20.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
certifi==2019.6.16=py37_1 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
entrypoints==0.3=py37_1000 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
widgetsnbextension==3.4.2=py37_1000 -> python[version='>=3.7,<3.8.0a0'] -> sqlite[version='>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']
testpath==0.4.2=py_1001 -> python -> sqlite[version='3.13.|3.20.|3.9.|>=3.20.1,<4.0a0|>=3.22.0,<4.0a0|>=3.23.1,<4.0a0|>=3.24.0,<4.0a0|>=3.25.1,<4.0a0|>=3.25.2,<4.0a0|>=3.25.3,<4.0a0|>=3.26.0,<4.0a0|>=3.27.2,<4.0a0|>=3.28.0,<4.0a0|>=3.29.0,<4.0a0']

Error while reproducing

Firstly, I really appreciate your great work!

While reproducing your work on the flickr30k dataset, an error occurred.
I think it is because I changed "train_flat.csv" to "train.csv" in ds_info.json, since I don't have train_flat.csv.
Please see the error and help me.

Error traceback:

Traceback (most recent call last):
File "/opt/pycharm/helpers/pydev/pydevd.py", line 1664, in
main()
File "/opt/pycharm/helpers/pydev/pydevd.py", line 1658, in main
globals = debugger.run(setup['file'], None, None, is_module)
File "/opt/pycharm/helpers/pydev/pydevd.py", line 1068, in run
pydev_imports.execfile(file, globals, locals) # execute the script
File "/opt/pycharm/helpers/pydev/_pydev_imps/_pydev_execfile.py", line 18, in execfile
exec(compile(contents+"\n", file, 'exec'), glob, loc)
File "/home/blackfoot/PycharmProjects/zsgnet-pytorch/code/main_dist.py", line 101, in
fire.Fire(main_dist)
File "/home/blackfoot/.local/lib/python3.6/site-packages/fire/core.py", line 138, in Fire
component_trace = _Fire(component, args, parsed_flag_args, context, name)
File "/home/blackfoot/.local/lib/python3.6/site-packages/fire/core.py", line 468, in _Fire
target=component.name)
File "/home/blackfoot/.local/lib/python3.6/site-packages/fire/core.py", line 672, in _CallAndUpdateTrace
component = fn(*varargs, **kwargs)
File "/home/blackfoot/PycharmProjects/zsgnet-pytorch/code/main_dist.py", line 89, in main_dist
learn = learner_init(uid, cfg)
File "/home/blackfoot/PycharmProjects/zsgnet-pytorch/code/main_dist.py", line 20, in learner_init
data = get_data(cfg)
File "/home/blackfoot/PycharmProjects/zsgnet-pytorch/code/dat_loader.py", line 237, in get_data
ds_name=ds_name, split_type='train')
File "/home/blackfoot/PycharmProjects/zsgnet-pytorch/code/dat_loader.py", line 84, in init
self.image_data = self._read_annotations(csv_file)
File "/home/blackfoot/PycharmProjects/zsgnet-pytorch/code/dat_loader.py", line 167, in _read_annotations
sample = trn_data['query'].iloc[0]
File "/home/blackfoot/.local/lib/python3.6/site-packages/pandas/core/indexing.py", line 1768, in getitem
return self._getitem_axis(maybe_callable, axis=axis)
File "/home/blackfoot/.local/lib/python3.6/site-packages/pandas/core/indexing.py", line 2138, in _getitem_axis
self._validate_integer(key, axis)
File "/home/blackfoot/.local/lib/python3.6/site-packages/pandas/core/indexing.py", line 2063, in _validate_integer
raise IndexError("single positional indexer is out-of-bounds")
IndexError: single positional indexer is out-of-bounds
Backend TkAgg is interactive backend. Turning interactive mode on.

Process finished with exit code 1

reproducing model on referit

hi, thank you for sharing codes .

I tried to reproduce a results on referit dataset, but i was not able to reach out within 40 epochs.

Acc was around 0.35 which was 0.2 lower than number in the paper.

I did not change any setting, so can you suggest any tips for reproducing?

My setting and logs are below.

ds_to_use: refclef
emb_dim: 300
epochs: 30
gamma: 2
lamb_reg: 1
load_normally: True
load_opt: False
local_rank: 0
lr: 0.0001
lstm_dim: 128
matching_threshold: 0.6
mdl_to_use: retina
num_gpus: 2
nw: 8
nwv: 8
only_test: False
only_val: False
opt_fn: Adam
opt_fn_params:
betas: [0.9, 0.99]
patience: 2
ratios: [1/2, 1, 2]
reduce_factor: 0.1
resize_img: [300, 300]
resume: True
resume_path:
scale_factor: 4
scales: [1, 2**(1/3), 2**(2/3)]
strict_load: True
tmp_path: ./tmp
use_bidirectional: True
use_focal: True
use_img: True
use_lang: True
use_multi: True
use_reduce_lr_plateau: True
use_same_atb: True
use_softmax: False
epochs trn_loss val_loss trn_cls_ls val_cls_ls trn_box_ls val_box_ls trn_Acc val_Acc trn_MaxPos val_MaxPos
11 0.6314 0.9759 0.6114 0.9393 0.0200 0.0366 0.4075 0.2945 0.9837 0.9692
12 0.5944 0.9765 0.5754 0.9421 0.0190 0.0345 0.4055 0.3128 0.9870 0.9684
13 0.5516 1.0506 0.5346 1.0120 0.0170 0.0387 0.4320 0.3136 0.9852 0.9610
14 0.5375 1.0342 0.5219 0.9972 0.0156 0.0371 0.4426 0.3063 0.9912 0.9653
15 0.3912 1.1772 0.3806 1.1373 0.0106 0.0399 0.5214 0.3515 0.9933 0.9601
16 0.3637 1.2997 0.3542 1.2566 0.0095 0.0431 0.5294 0.3480 0.9902 0.9522
17 0.3511 1.3821 0.3423 1.3358 0.0087 0.0463 0.5294 0.3515 0.9934 0.9463
18 0.3356 1.4874 0.3275 1.4390 0.0081 0.0484 0.5321 0.3502 0.9945 0.9409
19 0.3356 1.4966 0.3280 1.4483 0.0076 0.0484 0.5373 0.3507 0.9921 0.9398
20 0.3191 1.5148 0.3114 1.4659 0.0077 0.0489 0.5424 0.3529 0.9939 0.9395

the table 4 in paper.

Thank you for your great work. Recently, I am interested in the zero-shot setting for referring expression comprehension task and have read your paper ” Zero-Shot Grounding of Objects from Natural Language Queries".
I need do comparison experiment with your work, so I have a question about Table 4. The result is report on validation dataset or test dataset? I guest it for test dataset, right?

Thanks very much and looking forward to hearing from you.

dataset in table 5

Hi, could you release the dataset in Table5, thank you.

train time

Hi@TheShadow29,
How long does the paper use a single GPU for training on Flickr30K and Referit? I only have one Nvidia gpu:)

location of results.json file

In the data preparation script, I was not able to find the results.json file and the pipeline is not working because of that. I am not sure where exactly is the file present

which file should we download for the preparation of the Visual Genome datasets?

I would like to ask, in DATA_ README.md file, you mentioned downloading VG_100K, VG_100K_2 for the preparation of the Visual Genome datasets, but the web page https://visualgenome.org/api/v0/api_home.html, there are v1.0, v1.2 and v1.4. Which one should we download?

Problems in Unseen Splits

Thank you for your great work. Recently, I am interested in the zero-shot setting for referring expression comprehension task and have read your paper. I meet some problems when I reproduced the seen/unseen splits for Flickr30k Entities and Visual Genome datasets. Can you share the process you made the seen/unseen splits and the related codes? I want to know the exact phrases or objects you did the splits. Many thanks and looking forward to hearing from you.

Demo example

Hey! I love the project and I would like to use it in my master thesis (with the correspondent credit ofc), I managed to train a model but I feel a little bit lost when using it. Could it be possible to have a small example of loading a model and giving one image and one query and get the output?

Thank you so much in advanced,

Ferriol

problem while reproducing referit dataset

Thank you for your great work. I meet some problems when I reproduced the result on referit dataset. I use the same config with this file.
However, the validation accuracy is only 50% after 20 epochs while it should be 58% in the paper. I think maybe the config is different among different datasets. Looking forward to your help !

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.