mengyuanchen21 / cvpr2022-ftcl Goto Github PK
View Code? Open in Web Editor NEW[CVPR 2022] Fine-grained Temporal Contrastive Learning for Weakly-supervised Temporal Action Localization
License: MIT License
[CVPR 2022] Fine-grained Temporal Contrastive Learning for Weakly-supervised Temporal Action Localization
License: MIT License
After I downloaded and setting up the directory as instructed, I received an error by running the test:
python main_thu.py --test --checkpoint <home>/CVPR2022-FTCL/data/model/THUMOS-14/THUMOS_best.pth
Namespace(seed=0, model_name='default', group='default', checkpoint='<home>/CVPR2022-FTCL/data/model/THUMOS-14/THUMOS_best.pth', start_epoch=0, gpu='0', num_workers=0, dataset='THUMOS', batch_size=16, epochs=25, start_test_epoch=0, loss_lamb_lcs=0.1, loss_lamb_fsd=0.8, thres=0.92, lcs_len=30, fsd_len=5, fc_dim=1024, without_wandb=False, test=True, ftcl=False, without_lcs=False, without_fsd=False, class_name_lst=['BaseballPitch', 'BasketballDunk', 'Billiards', 'CleanAndJerk', 'CliffDiving', 'CricketBowling', 'CricketShot', 'Diving', 'FrisbeeCatch', 'GolfSwing', 'HammerThrow', 'HighJump', 'JavelinThrow', 'LongJump', 'PoleVault', 'Shotput', 'SoccerPenalty', 'TennisSwing', 'ThrowDiscus', 'VolleyballSpiking'], action_cls_num=20, dropout=0.7, lr=0.0001, weight_decay=5e-05, frames_per_sec=25, segment_frames_num=16, sample_segments_num=750, feature_dim=2048, cls_threshold=0.25, tiou_thresholds=array([0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9]), test_gt_file_path='<home>/CVPR2022-FTCL/data/THUMOS-14/gt.json', data_dir='<home>/CVPR2022-FTCL/data/THUMOS-14', test_upgrade_scale=20, nms_thresh=0.55, ins_topk_seg=8, con_topk_seg=3, bak_topk_seg=3, loss_lamb_1=0.002, loss_lamb_2=5e-05, loss_lamb_3=0.0002)
Traceback (most recent call last):
File "<home>/CVPR2022-FTCL/main_thu.py", line 146, in <module>
main(args)
File "<home>/CVPR2022-FTCL/main_thu.py", line 129, in main
test_dataset = build_dataset(args, phase="test", sample="uniform")
File "<home>/CVPR2022-FTCL/dataset/dataset_class.py", line 125, in build_dataset
return ACMDataset(args, phase, sample)
File "<home>/CVPR2022-FTCL/dataset/dataset_class.py", line 74, in __init__
self.data_list = list(open(os.path.join(self.data_dir, "split_test.txt")))
FileNotFoundError: [Errno 2] No such file or directory: '<home>/CVPR2022-FTCL/data/THUMOS-14/split_test.txt'
It seems the file split_test.txt
is missing, but I failed to find it in the provided google drive link or the THUMOS website. Can you help me out?
Thank you in advance!
Thank you for your contribution. Could you please tell me how to embed the results in the paper into existing methods? Is it mainly about introducing relevant loss functions?
1.I check one of the issues,which random seed you choose?
2.Besides, ACMnet origin repo asked 1000 epochs training, your pre-train model set only 500,is that enough?
3.Can i change the random seed on exist best model?
looking forward to your reply
my current best mAP 38
谢谢!
祝 生活愉快
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.