search

MVIG-SJTU / AlphaPose

Real-Time and Accurate Full-Body Multi-Person Pose Estimation&Tracking System
http://mvig.org/research/alphapose.html
Other
8.03k stars 1.97k forks source link

bad argument #2 to '?' (out of bounds #52

Closed shiyx1991 closed 6 years ago

shiyx1991 commented 6 years ago

I have tried several pictures in the demo without any problem . But when I want to use the demo on larger data, the program goes down and the error info:

0it [00:00, ?it/s] pose estimation with RMPE... MPII
/home/shiyx/torch/install/bin/luajit: bad argument #2 to '?' (out of bounds at/home/shiyx/torch/pkg/torch/lib/TH/generic/THStorage.c:202)

Fang-Haoshu commented 6 years ago

what is your command?

Fang-Haoshu commented 6 years ago

no response, close

tlatlbtle commented 6 years ago

The same issue.

Fang-Haoshu commented 6 years ago

we need more info

tlatlbtle commented 6 years ago

Command: $ ./run.sh --video /home/data/Men.mp4 --outdir examples/results/ --format cmu Message: 65 convert video to images... generating bbox from Faster RCNN... 2018-06-28 05:47:52.477516: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations. 2018-06-28 05:47:52.477575: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations. 2018-06-28 05:47:52.477598: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations. 2018-06-28 05:47:52.477615: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations. 2018-06-28 05:47:52.477641: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations. 2018-06-28 05:47:52.636950: I tensorflow/core/common_runtime/gpu/gpu_device.cc:940] Found device 0 with properties: name: Tesla P100-PCIE-16GB major: 6 minor: 0 memoryClockRate (GHz) 1.3285 pciBusID 051f:00:00.0 Total memory: 15.90GiB Free memory: 15.60GiB 2018-06-28 05:47:52.637009: I tensorflow/core/common_runtime/gpu/gpu_device.cc:961] DMA: 0 2018-06-28 05:47:52.637034: I tensorflow/core/common_runtime/gpu/gpu_device.cc:971] 0: Y 2018-06-28 05:47:52.637058: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1030] Creating TensorFlow device (/gpu:0) -> (device: 0, name: Tesla P100-PCIE-16GB, pci bus id: 051f:00:00.0) Loaded network ../output/res152/coco_2014_train+coco_2014_valminusminival/default/res152.ckpt /home/data/AlphaPose-RMPE-wjb/video-tmp

0it [00:00, ?it/s] pose estimation with RMPE... MPII /root/torch/install/bin/luajit: bad argument #2 to '?' (out of bounds at /root/torch/pkg/torch/lib/TH/generic/THStorage.c:202) stack traceback: [C]: at 0x7fe2f5c8bb40 [C]: in function '__index' /home/data/AlphaPose-RMPE-wjb/predict/util.lua:40: in function 'loadAnnotations' main-alpha-pose.lua:27: in main chunk [C]: in function 'dofile' /root/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk [C]: at 0x00405d50 parametric-pose-nms-MPII.py:104: UserWarning: loadtxt: Empty input file: "/home/data/AlphaPose-RMPE-wjb/examples/results/BBOX/score-proposals.txt" scores_proposals = np.loadtxt(os.path.join(outputpath,"BBOX/score-proposals.txt"), ndmin=1) parametric-pose-nms-MPII.py:18: UserWarning: loadtxt: Empty input file: "scores-proposals.txt" proposal_scores = np.loadtxt("scores-proposals.txt", ndmin=1)

tlatlbtle commented 6 years ago

Anyone help?

Fang-Haoshu commented 6 years ago

It seems the video is not converted to jpg correctly. Try mkdir examples/video-tmp ffmpeg -hide_banner -nostats -loglevel 0 -i ${VIDEO_FILE} -r 10 -f image2 examples/video-tmp"/%05d.jpg" and then run AlphaPose on the jpgs.

Our pytorch version will support video input soon. It's much faster and also accurate. Stay tuned!

tlatlbtle commented 6 years ago

Thx for your help! I run the command however get error due to wrong format of image name. It seems the command should be: ffmpeg -hide_banner -nostats -loglevel 0 -i ${VIDEO_FILE} -r 10 -f image2 examples/video-tmp"/%08d.jpg" So that it will meet the same format as: https://github.com/MVIG-SJTU/AlphaPose/blob/master/PoseFlow/alpha-pose-results-sample.json

tlatlbtle commented 6 years ago

I have modified the deepmatching.py and run it successfully.