buptxyb666 / adaptivepose Goto Github PK
View Code? Open in Web Editor NEWThis is an official implementation of our AAAI2022 paper AdaptivePose and Arxiv paper AdaptivePose++
License: MIT License
This is an official implementation of our AAAI2022 paper AdaptivePose and Arxiv paper AdaptivePose++
License: MIT License
Thanks for your work, I have a question,Do you mean "torch12 " is "pythoch1.12" ?Looking forward to your reply and answer
Hi, wheather estimate the three-dimensional coordinate information point of each person?
您好,请问通过bash main_hrnet48_coco640.sh,一开始训练就报错FileNotFoundError: [Errno 2] No such file or directory: '/home/aa/AdaptivePose-master/src/lib/../../exp/multi_pose_wodet/hrnet_48_coco640/model_last.pth'是什么原因,这不应该是在训练一轮后生成的么?训练的时候已经吧main_hrnet48_coco640.sh中的测试部分注释掉了。
Would author transplant code to the frame of mmpose ? So many start-of-the-art methods are published on mmpose~
当运行bash main_dla34_coco512.sh的时候,发生错误:test.py: error: unrecognized arguments: multi_pose_wodet
(我把训练部分删除了,只保留了test.py部分)
We found about 0.3-0.5 AP noise during training process. We think the user can decay the learning rate of the DCN in regression head to 1/10 for stabilizing the training.
当时使用torch==1.2.0和torchvision==0.4.0版本进行环境配置时,运行训练代码报错提示:CUDA error: CUBLAS_STATUS_EXECUTION_FAILED,猜测是因为CUDA与torch的版本对应不上;
但当我安装与CUDA11.6相对应的torch版本后,编译安装resample2d_cuda,又出现报错提示:undefined symbol: _ZN6caffe26detail37_typeMetaDataInstance_preallocated_32E
请问该如何解决?
Run the demo using command with model pose_hrnet_w48_384x288.pth
:
python demo.py multi_pose_wodet \
--gpus 0 \
--exp_id adp_hrnet_48 \
--dataset coco_hp_wodet \
--not_reg_offset \
--not_reg_hp_offset \
--K 20 \
--not_hm_hp \
--arch hrnet_48 \
--input_h 1920 \
--input_w 1080 \
--keep_res \
--debug 1 \
--demo ../data/video1/video1.mp4 \
--vis_thresh 0.1
and get results with bad pose estimation.
Input and output videos can be found here: BaiduDisk with code: hogg
Questions:
hrnet_48_crowdpose640.pth
?
PRETRAINED: 'hrnet_48_crowdpose640.pth'
in hrnet48.yaml
, but it did not work.hi, can you provide the pretrain_models of CrowdPose data, I train the model by pose_hrnet_w32_384x288.pth and the outcome is not great ,thank you!
Can you publish a py to onnx
你好 , 我运行demo.py的时候,Coco的数据格式,调试的时候adapt_pts 为空,然后报错,请问这个值是干嘛的? 怎么运行demo呢 谢谢!
由于本人远程实验室的机器,且没有root权限,所以无法使用您准备脚本配置环境,故均按照你的脚本改了改,现在在最后测试的时候出现这样的问题
python test.py multi_pose_wodet --exp_id coco512 --dataset coco_hp_wodet --resume --not_reg_offset --not_reg_hp_offset --K 20 --not_hm_hp --arch dla_34 --input_res 512 --keep_res
Keep resolution testing.
training chunk_sizes: [32]
The output will be saved to /home/renyouheng/AI_Code/AdaptivePose/src/lib/../../exp/multi_pose_wodet/coco512
heads {'hm': 1, 'hps': 34}
Namespace(K=20, aggr_weight=0.0, agnostic_ex=False, arch='dla_34', aug_ddd=0.5, aug_rot=0, batch_size=32, cat_spec_wh=False, center_thresh=0.1, chunk_sizes=[32], data_dir='/home/renyouheng/AI_Code/AdaptivePose/src/lib/../../data', dataset='coco_hp_wodet', debug=0, debug_dir='/home/renyouheng/AI_Code/AdaptivePose/src/lib/../../exp/multi_pose_wodet/coco512/debug', debugger_theme='white', demo='/home/users/yabo.xiao/CNet_v2/img', dense_hp=False, dense_wh=False, dep_weight=1, dim_weight=1, down_ratio=4, eval_oracle_dep=False, eval_oracle_hm=False, eval_oracle_hmhp=False, eval_oracle_hp_offset=False, eval_oracle_kps=False, eval_oracle_offset=False, eval_oracle_wh=False, exp_dir='/home/renyouheng/AI_Code/AdaptivePose/src/lib/../../exp/multi_pose_wodet', exp_id='coco512', fix_res=False, flip=0.5, flip_idx=[[1, 2], [3, 4], [5, 6], [7, 8], [9, 10], [11, 12], [13, 14], [15, 16]], flip_test=False, giou=False, giou_weight=1.0, gpus=[0], gpus_str='0', head_conv=256, heads={'hm': 1, 'hps': 34}, hide_data_time=False, hm_hp=False, hm_hp_weight=1, hm_weight=1, hp_weight=1, input_h=512, input_res=512, input_w=512, keep_res=True, kitti_split='3dop', load_model='/home/renyouheng/AI_Code/AdaptivePose/src/lib/../../exp/multi_pose_wodet/coco512/model_last.pth', lr=0.000125, lr_step=[90, 120], master_batch_size=32, mean=array([[[0.40789655, 0.44719303, 0.47026116]]], dtype=float32), metric='loss', mse_loss=False, nms=False, no_color_aug=False, norm_wh=False, not_cuda_benchmark=False, not_hm_hp=True, not_prefetch_test=False, not_rand_crop=False, not_reg_bbox=False, not_reg_hp_offset=True, not_reg_offset=True, num_classes=1, num_epochs=140, num_iters=-1, num_stacks=1, num_workers=4, off_weight=1, output_h=128, output_path='/opt/tiger/adaptivepose', output_res=128, output_w=128, pad=31, peak_thresh=0.2, print_iter=0, rect_mask=False, reg_bbox=True, reg_hp_offset=False, reg_loss='l1', reg_offset=False, resume=True, root_dir='/home/renyouheng/AI_Code/AdaptivePose/src/lib/../..', rot_weight=1, rotate=0, save_all=False, save_dir='/home/renyouheng/AI_Code/AdaptivePose/src/lib/../../exp/multi_pose_wodet/coco512', scale=0.4, scores_thresh=0.1, seed=317, shift=0.1, std=array([[[0.2886383 , 0.27408165, 0.27809834]]], dtype=float32), task='multi_pose_wodet', test=False, test_scales=[1.0], trainval=False, val_intervals=1, vis_thresh=0.3, wh_weight=0.1)
==> initializing coco 2017 val data.
loading annotations into memory...
Done (t=0.35s)
creating index...
index created!
Loaded val 5000 samples
Creating model...
backbone_Params: 19.72M
Traceback (most recent call last):
File "test.py", line 134, in
prefetch_test(opt)
File "test.py", line 58, in prefetch_test
detector = Detector(opt)
File "/home/renyouheng/AI_Code/AdaptivePose/src/lib/detectors/multi_pose_wodet.py", line 26, in init
super(MultiPoseDetector_wodet, self).init(opt)
File "/home/renyouheng/AI_Code/AdaptivePose/src/lib/detectors/base_detector.py", line 24, in init
self.model = create_model(opt.arch, opt.heads, opt.head_conv, is_train=False)
File "/home/renyouheng/AI_Code/AdaptivePose/src/lib/models/model.py", line 30, in create_model
model = get_model(num_layers=num_layers, heads=heads, head_conv=head_conv)
File "/home/renyouheng/AI_Code/AdaptivePose/src/lib/models/networks/pose_dla_dcn_v2.py", line 562, in get_pose_net
head_conv=head_conv)
File "/home/renyouheng/AI_Code/AdaptivePose/src/lib/models/networks/pose_dla_dcn_v2.py", line 528, in init
self.sampler = Feat_sampler(head_conv)
TypeError: init() missing 1 required positional argument: 'hps_channel'
不知道demo怎么运行起来
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.