Giter VIP home page Giter VIP logo

ucas-vg / pointtinybenchmark Goto Github PK

View Code? Open in Web Editor NEW
649.0 22.0 80.0 47.1 MB

Point based and tiny object detection and localization code set of UCAS-VG

Home Page: http://vision.ucas.ac.cn/sources

License: MIT License

Python 50.46% Shell 0.22% Jupyter Notebook 49.27% Dockerfile 0.03% Makefile 0.01% Batchfile 0.01%
detection tiny-object scale-match benchmark dataset eccvw localization point-supervision

pointtinybenchmark's Introduction

PointTinyBenchmark


PointTinyBenchmark is an open source toolbox for object localization and detection tasks on top of mmdetection.

To date, It implements the following benchmark and algorithms:

News

Others

Citation

And if the following works do some help for your research, please cite:

@inproceedings{CPR,
  author    = {Yu, Xuehui and Chen, Pengfei and Wu, Di and Hassan, Najmul and Li,
Guorong and Yan, Junchi and Shi, Humphrey and Ye, Qixiang and Han, Zhenjun},
  title     = {Object Localization under Single Coarse Point Supervision},
  booktitle = {CVPR},
  year      = {2022},
}
@article{yu20201st,
  title={The 1st Tiny Object Detection Challenge: Methods and Results},
  author={Yu, Xuehui and Han, Zhenjun and Gong, Yuqi and Jan, Nan and Zhao, Jian and Ye, Qixiang and Chen, Jie and Feng, Yuan and Zhang, Bin and Wang, Xiaodi and others},
  journal={arXiv preprint arXiv:2009.07506},
  year={2020}
}
@inproceedings{yu2020scale,
  title={Scale Match for Tiny Person Detection},
  author={Yu, Xuehui and Gong, Yuqi and Jiang, Nan and Ye, Qixiang and Han, Zhenjun},
  booktitle={The IEEE Winter Conference on Applications of Computer Vision},
  pages={1257--1265},
  year={2020}
}
@inproceedings{P2BNet,
  author    = {Pengfei Chen, Xuehui Yu, Xumeng Han, Najmul Hassan, Kai Wang, Jiachen
Li, Jian Zhao, Humphrey Shi, Zhenjun Han, and Qixiang Ye},
  title     = {Point-to-Box Network for Accurate Object Detection via Single Point Supervision},
  booktitle = {ECCV},
  year      = {2022},
}
@inproceedings{SSDDET,
  author    = {Wu, Di and Chen, Pengfei and Yu, Xuehui and Li,
Guorong and Han, Zhenjun and Jiao, Jianbin},
  title     = {Spatial Self-Distillation for Object Detection with Inaccurate Bounding Boxes},
  booktitle = {ICCV},
  year      = {2023},
}

pointtinybenchmark's People

Contributors

darren-pfchen avatar dependabot[bot] avatar yinglang avatar zhenjunh avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

pointtinybenchmark's Issues

Number of image ids is over 8000

Hi
I'm using COCO to load this file "tiny_set/erase_with_uncertain_dataset/annotations/corner/task/tiny_set_train_sw640_sh512_all.json"

Then if I get the image ids, then I have len(img_ids) = 8256

img_ids = coco.getImgIds()

I thought we only have around 793 images in train?

Scale Match offline

I saw a few question related to scale matching for custom datasets but I did now see any clear answer. My question is related to Coco dataset scale matching. I want to coco annotations and images to match my custom dataset. Then, I will train yolov4(not mask rcnn) over coco and then will use coco weights to fine tune network over my custom dataset. Can you please provide information about coco annotation conversion for custom dataset? Btw, thank you for your works...

how to set configs?

CUDA_VISIBLE_DEVICES=0,1 python -m torch.distributed.launch --nproc_per_node=$NGPUS tools/train_test_net.py --config ${config_path}
how to set ''--config ${config_path}''

relative size

Hello, how is the relative size of the object in Figure 1 calculated? I divided the target area by the area of the picture for statistics, and the result obtained is very different from yours.

ImportError: DLL load failed: 找不到指定的程序

Excuse me, I encountered this error while running the program, how should I solve it?Did the compilation fail?
Traceback (most recent call last):
File "D:/BaiduNetdiskDownload/tiny_face/TinyBenchmark-master/tiny_benchmark/tools/train_test_net.py", line 19, in
from maskrcnn_benchmark.engine.inference import inference
File "D:\Users\86183\Anaconda3\envs\maskrcnn_benchmark\lib\site-packages\maskrcnn_benchmark-0.1-py3.7-win-amd64.egg\maskrcnn_benchmark\engine\inference.py", line 10, in
from tiny_benchmark.maskrcnn_benchmark.data.datasets.evaluation import evaluate
File "D:\BaiduNetdiskDownload\tiny_face\TinyBenchmark-master\tiny_benchmark\maskrcnn_benchmark\data\datasets\evaluation_init_.py", line 3, in
from .coco import coco_evaluation
File "D:\BaiduNetdiskDownload\tiny_face\TinyBenchmark-master\tiny_benchmark\maskrcnn_benchmark\data\datasets\evaluation\coco_init_.py", line 1, in
from .coco_eval import do_coco_evaluation
File "D:\BaiduNetdiskDownload\tiny_face\TinyBenchmark-master\tiny_benchmark\maskrcnn_benchmark\data\datasets\evaluation\coco\coco_eval.py", line 10, in
from maskrcnn_benchmark.structures.boxlist_ops import boxlist_iou
File "D:\Users\86183\Anaconda3\envs\maskrcnn_benchmark\lib\site-packages\maskrcnn_benchmark-0.1-py3.7-win-amd64.egg\maskrcnn_benchmark\structures\boxlist_ops.py", line 6, in
from tiny_benchmark.maskrcnn_benchmark.layers import nms as box_nms
File "D:\BaiduNetdiskDownload\tiny_face\TinyBenchmark-master\tiny_benchmark\maskrcnn_benchmark\layers_init
.py", line 9, in
from .nms import nms
File "D:\BaiduNetdiskDownload\tiny_face\TinyBenchmark-master\tiny_benchmark\maskrcnn_benchmark\layers\nms.py", line 3, in
from maskrcnn_benchmark import _C
ImportError: DLL load failed: 找不到指定的程序。

IOD

image

Hello, does this result of your paper use IOD?

scale match

DATASETS:
TRAIN: ("coco_2014_train", "coco_2014_valminusminival", "coco_2014_minival")
TEST: ("coco_2014_minival",)

DATALOADER:
SIZE_DIVISIBILITY: 32
USE_MORE_DA: 4
USE_SCALE_MATCH: True
SCALE_MATCH:
TYPE: 'MonotonicityScaleMatch'
SOURCE_ANNO_FILE: '${YOUR_DATASET_DIR}/coco/annotations/instances_simple_merge2014.json'
TARGET_ANNO_FILE: '${YOUR_DATASET_DIR}/tiny_set/erase_with_uncertain_dataset/annotations/corner/task/tiny_set_train_sw640_sh512_all.json'
BINS: 100
DEFAULT_SCALE: 0.25
SCALE_RANGE: (0.1, 1.)
OUT_SCALE_DEAL: 'clip'、

Hello,I want use scale match ,can I know source_anno_file is the annotation of train_set?Or the annotation of train_set still adopts the original annotation of coco dataset

pretrain-model

FileNotFoundError: [Errno 2] No such file or directory: '/home/hui/params/pretrained_model/ImageNetPretrained/MSRA/R-50.pkl'
can you share me a pretrain model?

How to test using pre-trained weights and my test images

Hey, firstly, I'm new to maskrcnn. I'm trying to used pre-trained weights to infer tiny persons on an image, but I am not sure how to do it.
I tried to follow demo, but it uses coco_demo.run_on_opencv_image(image) which I'm confused how to adapt for the inference used in test_net.py

What I have tried:

  1. Testing on coco_val2014 dataset. I downloaded it, and put it inside dataset/coco along with instances_minival2014.json file.
    1.1 I added a new config in path_config.py with:
coco_val_test: {
   img_dir: #path to coco dataset images,
   ann_dir: #path to `instances_minival2014`
}

1.2 Updated this config file to include coco_val_test under test, and updated WEIGHT to #path to params/model_only_e2e_faster_rcnn_R_50_FPN_1x.pth downloaded from Drive
1.3 Ran CUDA_VISIBLE_DEVICES=0,1 python -m torch.distributed.launch --nproc_per_node=$NGPUS tools/test_net.py --config ${config_path}

It ran through and generated a bbox.json. I tried to put these boxes on an image, but these boxes didn't make any sense (incorrect drawings).

Similarly, I tried making my own version of instances_minival2014.json with only 1 image. It also ran the test_net.py but the output bbox is not coming correct.

Looks like this:
image

Can some one please help me set it up? Any help will be appreciated!

AP gap between TinyBenchmark and MMDetection

Hello, I appreciate your work and I tried to re-implement the baseline with FPN on MMDetection, but the AP50tiny metric is extremely low (only about 7%) with the same settings on MMDetection. Has anyone ever come across the same problem and could you please give me some suggestions about the parameter setting of models or some other details?

the yaml and the pth

I downloaded the models, but how do they correspond to yaml in config,thanks a lot

Test and display picture

Can I test only a single picture and display the picture of the detection result?
Can you share with me the test code for a single picture?thanks!

Code license

I want you to tell me the license.
I'm happy at MIT, but I want you to set the appropriate license you think.

how to train

Do I need to download the coco dataset, and if necessary, which file directory should I put it in

Some question about MSM

I use an extra dataset to match a target datasets with MSM. The extra dataset has 17 categories and the target dataste has 3 categories. Before MSM,the extra datasets have dissimilar distributions with target datasets. After MSM,the extra datasets have similar distributions with target dataset. In order to ensure the consistency of the image scale, I put the extra image in the upper left corner, and fill in the remaining positions with 0.
But when I use MSM extra dataset as pre-trained model to train target dataset, it can only speed up the convergence, but it cannot improve performance. Do you have some suggestions. The experiment is based on mmdetection.

run with one gpu?

AssertionError: NUM_GPU is not equal to visible GPU count 2 vs 1.
i just have one gpu, how to solve this problem? thankyou

lack of a file of json

when i train with the config file in TINYPERSON/FPN/BASELINE1 the e2e_faster_rcnn_R_50_FPN_1x_cocostyle_baseline1.yaml

it works for a while,later it stop saying:
No such file or directory: '/media/ubuntu/0000BFD200076E00/U-net/TinyBenchmark-master/dataset/tiny_set/annotations/corner/task/tiny_set_test_sw640_sh512_all.json'

i check the folder ,it does lack the file ,only/tiny_set_train_sw640_sh512_all.json' why

RuntimeError: Error compiling objects for extension

While installing , i got the following error ,,

running build running build_py running build_ext building 'maskrcnn_benchmark._C' extension Emitting ninja build file /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/build.ninja... Compiling objects... Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N) [1/7] /usr/local/cuda/bin/nvcc -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/nms.cu -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/nms.o -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr --compiler-options '-fPIC' -DCUDA_HAS_FP16=1 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_70,code=sm_70 -std=c++14 FAILED: /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/nms.o /usr/local/cuda/bin/nvcc -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/nms.cu -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/nms.o -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr --compiler-options '-fPIC' -DCUDA_HAS_FP16=1 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_70,code=sm_70 -std=c++14 /bin/sh: 1: /usr/local/cuda/bin/nvcc: Permission denied [2/7] /usr/local/cuda/bin/nvcc -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/SigmoidFocalLoss_cuda.cu -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/SigmoidFocalLoss_cuda.o -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr --compiler-options '-fPIC' -DCUDA_HAS_FP16=1 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_70,code=sm_70 -std=c++14 FAILED: /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/SigmoidFocalLoss_cuda.o /usr/local/cuda/bin/nvcc -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/SigmoidFocalLoss_cuda.cu -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/SigmoidFocalLoss_cuda.o -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr --compiler-options '-fPIC' -DCUDA_HAS_FP16=1 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_70,code=sm_70 -std=c++14 /bin/sh: 1: /usr/local/cuda/bin/nvcc: Permission denied [3/7] /usr/local/cuda/bin/nvcc -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/ROIPool_cuda.cu -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/ROIPool_cuda.o -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr --compiler-options '-fPIC' -DCUDA_HAS_FP16=1 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_70,code=sm_70 -std=c++14 FAILED: /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/ROIPool_cuda.o /usr/local/cuda/bin/nvcc -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/ROIPool_cuda.cu -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/ROIPool_cuda.o -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr --compiler-options '-fPIC' -DCUDA_HAS_FP16=1 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_70,code=sm_70 -std=c++14 /bin/sh: 1: /usr/local/cuda/bin/nvcc: Permission denied [4/7] /usr/local/cuda/bin/nvcc -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/ROIAlign_cuda.cu -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/ROIAlign_cuda.o -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr --compiler-options '-fPIC' -DCUDA_HAS_FP16=1 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_70,code=sm_70 -std=c++14 FAILED: /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/ROIAlign_cuda.o /usr/local/cuda/bin/nvcc -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/ROIAlign_cuda.cu -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cuda/ROIAlign_cuda.o -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr --compiler-options '-fPIC' -DCUDA_HAS_FP16=1 -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_70,code=sm_70 -std=c++14 /bin/sh: 1: /usr/local/cuda/bin/nvcc: Permission denied [5/7] c++ -MMD -MF /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/vision.o.d -pthread -B /home/sa/anaconda3/envs/maskrcnn_benchmark/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/vision.cpp -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/vision.o -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++14 FAILED: /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/vision.o c++ -MMD -MF /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/vision.o.d -pthread -B /home/sa/anaconda3/envs/maskrcnn_benchmark/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/vision.cpp -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/vision.o -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++14 cc1plus: warning: command line option ‘-Wstrict-prototypes’ is valid for C/ObjC but not for C++ cc1plus: error: /usr/local/cuda/include: Permission denied [6/7] c++ -MMD -MF /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/ROIAlign_cpu.o.d -pthread -B /home/sa/anaconda3/envs/maskrcnn_benchmark/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/ROIAlign_cpu.cpp -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/ROIAlign_cpu.o -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++14 FAILED: /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/ROIAlign_cpu.o c++ -MMD -MF /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/ROIAlign_cpu.o.d -pthread -B /home/sa/anaconda3/envs/maskrcnn_benchmark/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/ROIAlign_cpu.cpp -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/ROIAlign_cpu.o -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++14 cc1plus: warning: command line option ‘-Wstrict-prototypes’ is valid for C/ObjC but not for C++ cc1plus: error: /usr/local/cuda/include: Permission denied [7/7] c++ -MMD -MF /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/nms_cpu.o.d -pthread -B /home/sa/anaconda3/envs/maskrcnn_benchmark/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/nms_cpu.cpp -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/nms_cpu.o -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++14 FAILED: /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/nms_cpu.o c++ -MMD -MF /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/nms_cpu.o.d -pthread -B /home/sa/anaconda3/envs/maskrcnn_benchmark/compiler_compat -Wl,--sysroot=/ -Wsign-compare -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -DWITH_CUDA -I/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/TH -I/home/sa/anaconda3/envs/maskrcnn_benchmark/lib/python3.8/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/sa/anaconda3/envs/maskrcnn_benchmark/include/python3.8 -c -c /sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/nms_cpu.cpp -o /sa//TinyBenchmark/tiny_benchmark/build/temp.linux-x86_64-3.8/sa//TinyBenchmark/tiny_benchmark/maskrcnn_benchmark/csrc/cpu/nms_cpu.o -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_C -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++14 cc1plus: warning: command line option ‘-Wstrict-prototypes’ is valid for C/ObjC but not for C++ cc1plus: error: /usr/local/cuda/include: Permission denied ninja: build stopped: subcommand failed.

I have corrected the error by 'sudo python setup.py build develop'
Now getting the following error, I have installed pytorch using the command ..

File "setup.py", line 7, in <module>
  import torch
ImportError: No module named torch

ImportError: cannot import name 'inference_trainer'

Hi, I meet a question when I run train_test_net.py
just like this:

Traceback (most recent call last):
File "/home/zhang/TinyBenchmark/tiny_benchmark/tools/train_test_net.py", line 20, in
from maskrcnn_benchmark.engine import inference_trainer
ImportError: cannot import name 'inference_trainer'

My install is right, and inference_trainer.py is exist.

test annotation

Do you have test annotation? or do you have a test set that split a sub-set from training ?

Scale Match Failed

When I modify the config file and train on my own dataset, this error happens:
File "tiny_benchmark/maskrcnn_benchmark/data/datasets/ship.py", line 110, in getitem
img, target = self.transforms(img, target)
File "maskrcnn_benchmark/data/transforms/transforms.py", line 13, in call
image, target = t(image, target)
File "maskrcnn_benchmark/data/transforms/scale_match.py", line 293, in call
dst_size = self._sample_by_distribute(src_size)
File "maskrcnn_benchmark/data/transforms/scale_match.py", line 255, in _sample_by_distribute
bin_i = np.nonzero(src_size <= self.src_sizes[1:] + 1e-6)[0][0]
IndexError: index 0 is out of bounds for axis 0 with size 0

I have tried modify the bins number amd the scale range, but it failed.
Could you give some suggestions ?

Training my own dataset

Hello, I think this is a nice work.
I have a question, can this Tinybenchmark use my own data set to detect the target I need?

For example, I want to detect different types of cars (there are probably 4 types of targets), but they are all very small targets.

What do I need to change? Can you give me some suggestions? Thank you!

A problem when I run evaluate_tiny.py !!!

When I run TinyBenchmark/tiny_benchmark/MyPackage/tools/evaluate/ #evaluate_tiny.py, I got an error:

/media/hp208/4t/zhaoxingjie/project_graduation/TinyBenchmark/venv/bin/python /media/hp208/4t/zhaoxingjie/project_graduation/TinyBenchmark/tiny_benchmark/MyPackage/tools/evaluate/evaluate_tiny.py
Traceback (most recent call last):
  File "/media/hp208/4t/zhaoxingjie/project_graduation/TinyBenchmark/tiny_benchmark/MyPackage/tools/evaluate/evaluate_tiny.py", line 14, in <module>
    from MyPackage.visulize.plot_train_log import parse_log, replace_key
ImportError: No module named 'MyPackage.visulize'

Anyone meet the same problem?

some question about tiny cityperson

Hi, sorry to disturb you.
Could you tell me more details about training tiny citypersons?

1. Used several categories

There are four categories in Citypersons paper. How many categories did you use?
image

2. How to get the data in paths_catalog.py?

I only know that tiny citypersons downsample cityperson to 4*4. But I don’t know if tiny citypersons also use the same crop and erasure as tinyperson. Or you can provide training and val data for tiny citypersons, thank you very much!
image

File path problem

i have modified path in ${TinyBenchmark}/tiny_benchmark/maskrcnn_benchmark/config/paths_catalog.py
But there are still path problems,What should I do?
Uploading QQ图片20200428233053.png…
[error]:No such file or directory:
'/home/$user/TinyBenchmark/dataset/tiny_set/erase_with_uncertain_dataset/annotations/corner/task/tiny_set_train_sw640_sh512_all.json'

Data annotation problems

Hi! When I make my own dataset, the label of the object needs to be like instance segmentation, or just bbox (x1, y1, x2, y2).
Because I see that your label has both "segmentation" and "bbox". Would you like to use the "segmentation" label during training?

The result of coco pretrained model is very poor

Hi, I'm sorry to disturb you.
When I train to get the coco pretrained model, I get very poor results.

Just like this:

coco results

my training command:

export NGPUS=4
CUDA_VISIBLE_DEVICES=4,5,6,7 python -m torch.distributed.launch --nproc_per_node=$NGPUS tools/train_test_net.py

and my yaml file is:

MODEL:
META_ARCHITECTURE: "GeneralizedRCNN"
WEIGHT: "catalog://ImageNetPretrained/MSRA/R-50"
BACKBONE:
CONV_BODY: "R-50-FPN"
RESNETS:
BACKBONE_OUT_CHANNELS: 256
RPN:
USE_FPN: True
ANCHOR_STRIDE: (4, 8, 16, 32, 64)
ANCHOR_SIZES: (8.31, 12.5, 18.55, 30.23, 60.41) # adjust here
ASPECT_RATIOS: (0.5, 1.3, 2) # adjust here
PRE_NMS_TOP_N_TRAIN: 2000
PRE_NMS_TOP_N_TEST: 1000
POST_NMS_TOP_N_TEST: 1000
FPN_POST_NMS_TOP_N_TEST: 1000
ROI_HEADS:
USE_FPN: True
ROI_BOX_HEAD:
POOLER_RESOLUTION: 7
POOLER_SCALES: (0.25, 0.125, 0.0625, 0.03125)
POOLER_SAMPLING_RATIO: 2
FEATURE_EXTRACTOR: "FPN2MLPFeatureExtractor"
PREDICTOR: "FPNPredictor"
INPUT:
USE_SCALE: True
SCALES: (0.25,) # here only used for test dataset transform, when scale match on
DATASETS:
TRAIN: ("coco_2014_train", "coco_2014_valminusminival", "coco_2014_minival")
TEST: ("coco_2014_minival",)
DATALOADER:
SIZE_DIVISIBILITY: 32
USE_MORE_DA: 4
USE_SCALE_MATCH: True
SCALE_MATCH:
TYPE: 'MonotonicityScaleMatch'
SOURCE_ANNO_FILE: '/home/zhangyc/datasets/coco/annotations/instances_simple_merge2014.json'
TARGET_ANNO_FILE: '/home/zhangyc/datasets/Tiny/train/annotations/train.json'
BINS: 100
DEFAULT_SCALE: 0.25
SCALE_RANGE: (0.1, 1.)
OUT_SCALE_DEAL: 'clip'

SOLVER:
WARMUP_METHOD: "constant"
TEST_ITER: 10000
CHECKPOINT_PERIOD: 30000
BASE_LR: 0.02
WEIGHT_DECAY: 0.0001
STEPS: (60000, 80000)
MAX_ITER: 90000
IMS_PER_BATCH: 16
NUM_GPU: 4

TEST:
IMS_PER_BATCH: 16

OUTPUT_DIR: 'outputs/tiny_coco/FPN_msm_tiny_b4'

and my training loss is also high:

loss1

loss

my log.txt is: log.txt

question:

May I ask if my parameters or training methods are wrong.
Can you give me some suggestion? Thanks!
Look forward to your reply!

Low results on validation set

Hi, I test my model on all train images (without corner) and got 11.1 map on tiny 50. However, the same model I submitted to the challenge website on the test set can achieve 50 map on tiny 50. So, would you share you your config which test on the test or train set without corner info (no need for merge file)

test conf thres

Hi, sorry to disturb you.
What is the confidence threshold used in the test time? I did not see this setting in the .yaml file.
may be 0.01 or 0.001 ?

About Change of Evaluation

Hi. I have some questions about your paper.

  1. In your paper
    image

AP_tiny_50 = 43.55 for detector (Faster RCNN), AP_tiny_50 = 47.29 when use MSM COCO pretrained.
"AP_tiny_50 = 43.55 for detector" does it mean that the original network is used for training instead of the MSM method?

  1. In your github
    image
    image

AP_tiny_50 = 47.35 for detector (Faster RCNN), AP_tiny_50 = 51.33 when use SM COCO pretrained.
"AP_tiny_50 = 47.35 for detector" does it mean that the original network is used for training instead of the SM method?

  1. IOD

Do "AP_tiny_50 = 43.55 for detector" and "AP_tiny_50 = 47.35 for detector" both use the original Faster RCNN (not use SM or MSM)? Is the difference between them in the evaluation method? The latter uses the IOD method for evaluation?

image

Does "AP_tiny_50 = 43.55 for detector" use IOD, and "AP_tiny_50 = 47.35 for detector" does not use IOD?

image
But it is mentioned in your paper that "AP_tiny_50 = 47.29" seems to use IOD.

Is there any difference between these two sets of experiments?
Sorry to disturb you and look forward to your reply, thank you!

Import Error while install from pip

hi~, firstly, I think it is a good job.
However, when I run your code , I got the following error:
ImportError: /home/ubuntub/Guo/TinyBenchmark-master/tiny_benchmark/maskrcnn_benchmark/_C.cpython-35m-x86_64-linux-gnu.so: undefined symbol: _ZN6caffe28TypeMeta21_typeMetaDataInstanceIN3c108BFloat16EEEPKNS_6detail12TypeMetaDataEv

But I have done python setup.py build develop following your instructions.
I am sincerely looking forword to your reply.
Thanks!

pth and the yaml

I downloaded the models, but how do these models correspond to yaml in config? Can you provide a demo to use it? I'm a beginner, thank you

some training question

Hi!I'm sorry to disturb you!
In your paper, you used the coco dataset to pretrain and SM to the tinyperson dataset. How can I get the coco pretrained weight of my own dataset? And I still don't know how to use my own dataset to train?
I‘m looking forward to your reply! Thank you!

COCOPretrained

is models in params/cocopretrained only trained on coco? or is trained on tiny person initialied by mscoco pretrained params?

Recommend Projects

  • React photo React

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

  • Vue.js photo Vue.js

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

  • Typescript photo Typescript

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

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo 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.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

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

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.