![](https://img-blog.csdnimg.cn/20190918140145169.png?x-oss-process=image/resize,m_fixed,h_224,w_224)
目标跟踪
目标跟踪
朱朱呀
这个作者很懒,什么都没留下…
展开
-
dimp18训练结果
[train: 97, 526 / 1000] FPS: 16.8 (5.8) , Loss/total: 0.43429 , Loss/iou: 0.02153 , Loss/target_clf: 0.06703 , Loss/test_init_clf: 0.07536 , Loss/test_iter_clf: 0.27037 , ClfTrain/test_loss: 0.00067 , ClfTrain/test_init_loss: 0.00075 , ClfT原创 2020-09-05 12:42:25 · 314 阅读 · 1 评论 -
PrDiMP_resnet50(修改学习率)训练记录
run_training.py dimp prdimp50batchsize由10改成6 不然运行不起来但是只减小batchsize训练的效果不好,loss不是最优的,故同时减小学习率,如下所示。结果保存在result_model_prdimp_delr原始学习率 optimizer = optim.Adam([{'params': actor.net.classifier.parameters(), 'lr': 1e-3}, .原创 2020-08-29 15:07:24 · 2293 阅读 · 0 评论 -
PrDiMP_resnet50训练记录
run_training.py dimp prdimp50batchsize由10改成6 不然运行不起来结果保存在result_model_prdimp[train: 1, 1 / 4333] FPS: 0.6 (0.6) , Loss/total: 6.31510 , Loss/bb_ce: 4.56391 , ClfTrain/clf_ce: 3.74005[train: 1, 2 / 4333] FPS: 1.2 (8.6) , Loss/total: 5.88499 ,.原创 2020-08-24 20:10:03 · 372 阅读 · 0 评论 -
PrDiMP18(epoch=90)在UAV123的测试结果
使用prdimp resnet18 batch_size/train_epoch=16/90测试结果保存在tracking_results_me_prdimp18_epoch90/result_plots_me_prdimp18_epoch90Running: myexperiments dimp_uavEvaluating 1 trackers on 123 sequencesTracker: dimp prdimp18 0 , Sequence: uav_bike1/hom.原创 2020-08-23 13:04:49 · 427 阅读 · 1 评论 -
PrDiMP18(epoch=90)在OTB100的测试结果
使用prdimp resnet18 batch_size/train_epoch=16/90测试结果保存在tracking_results_me_prdimp18_epoch90/result_plots_me_prdimp18_epoch90Running: myexperiments dimp_otbEvaluating 1 trackers on 98 sequencesTracker: dimp prdimp18 0 , Sequence: Basketball/ho.原创 2020-08-23 13:00:31 · 357 阅读 · 1 评论 -
PrDiMP18(epoch=50)在UAV123的测试结果
使用prdimp resnet18 batch_size/train_epoch=16/50测试结果保存在tracking_results_me_prdimp18/result_plots_me_prdimp18Running: myexperiments dimp_uavEvaluating 1 trackers on 123 sequencesTracker: dimp prdimp18 0 , Sequence: uav_bike1/home/zhuzhu/anacond.原创 2020-08-23 12:52:52 · 434 阅读 · 1 评论 -
ATOM resnet50/PrDiMP18 在UAV123数据集测试结果
ATOM resnet50/PrDiMP18(epoch=90) 在UAV123数据集测试结果(tracker_num=1)结果保存在tracking_results_me_prdimp18_epoch90_atom5/result_plots_me_prdimp18_epoch90_atom50Running: myexperiments atom_dimp_uavEvaluating 2 trackers on 123 sequencesTracker: dimp prdimp.原创 2020-08-23 11:48:09 · 533 阅读 · 0 评论 -
ATOM resnet18/PrDiMP18在UAV123/OTB100数据集测试结果
ATOM resnet18/PrDiMP18(epoch=90) 在UAV123数据集测试结果(tracker_num=1)结果保存在tracking_results_me_prdimp18_epoch90_atom18/result_plots_me_prdimp18_epoch90_atom18Running: myexperiments atom_dimp_uavEvaluating 2 trackers on 123 sequencesTracker: dimp prdim.原创 2020-08-23 11:30:33 · 636 阅读 · 6 评论 -
ATOM resnet50/PrDiMP18 在OTB100数据集测试结果
ATOM resnet50/PrDiMP18 在OTB100数据集测试结果(tracker_num=1)结果保存在tracking_results_me_prdimp18_epoch90_atom5/result_plots_me_prdimp18_epoch90_atom50Running: myexperiments atom_dimp_otbEvaluating 2 trackers on 98 sequencesTracker: dimp prdimp18 0 , ..原创 2020-08-23 11:15:13 · 702 阅读 · 2 评论 -
ATOM_resnet50在OTB100数据集测试结果
ATOM resnet50 在OTB100数据集测试结果(tracker_num=1)结果保存在tracking_results_me_resnet50/result_plots_me_resnet50原创 2020-08-22 19:22:26 · 967 阅读 · 1 评论 -
ATOM_resnet50在UAV123数据集测试结果
ATOM resnet50 在OTB100数据集测试结果(tracker_num=3)结果保存在tracking_results_me_resnet50/result_plots_me_resnet50Running: myexperiments atom_uavEvaluating 3 trackers on 123 sequencesTracker: atom default 0 , Sequence: uav_bike1/home/zhuzhu/anaconda3/env.原创 2020-08-22 19:05:52 · 800 阅读 · 1 评论 -
PrDiMP18在OTB100的测试结果
使用prdimp resnet18 batch_size/train_epoch=16/50测试结果保存在tracking_results_me_prdimp18/result_plots_me_prdimp18Running: myexperiments dimp_otbEvaluating 1 trackers on 98 sequencesTracker: dimp prdimp18 0 , Sequence: Basketball/home/zhuzhu/anacon.原创 2020-08-22 18:54:31 · 458 阅读 · 0 评论 -
PrDiMP_resnet18在运行第42个epoch时报错
[train: 42, 608 / 1625] FPS: 12.0 (24.0) , Loss/total: 4.10781 , Loss/bb_ce: -4.89841 , ClfTrain/clf_ce: 2.74648[train: 42, 609 / 1625] FPS: 11.9 (2.2) , Loss/total: 4.10778 , Loss/bb_ce: -4.89871 , ClfTrain/clf_ce: 2.74647[train: 42, 610 / 1原创 2020-08-21 13:18:08 · 369 阅读 · 3 评论 -
ATOM_resnet18_otb100数据集 用自己训练的模型进行测试
run_experiment.py debug=0 atom default atom_otb 结果保存在pytracking.tracking_results_me.atom.default_000Running: myexperiments atom_otbEvaluating 1 trackers on 98 sequencesTracker: atom default 0 , Sequence: Basketball/home/zhuzhu/anaconda3/envs.原创 2020-08-20 15:51:16 · 1284 阅读 · 6 评论 -
PrDiMP_resnet18训练记录
run_training.py dimp prdimp18结果保存在result_model_prdimpTraining: dimp prdimp18[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45.原创 2020-08-19 19:53:12 · 343 阅读 · 0 评论 -
ATOM_resnet18_uav数据集 用自己训练的模型进行测试
run_experiment.py debug=0 atom default 结果保存在pytracking.tracking_results_me.atom.default_000Running: myexperiments uav_testEvaluating 1 trackers on 123 sequencesTracker: atom default 0 , Sequence: uav_bike1/home/zhuzhu/anaconda3/envs/pytrackin.原创 2020-08-18 20:39:20 · 837 阅读 · 2 评论 -
pytracking-ATOM代码--用自己训练的模型进行跟踪
把ltr.result_model.checkpoints.ltr.bbreg.atom下的文件ATOMnet_ep0050pth.tar移动到pytracking.networks下,并改名去掉.tar后缀(即解压);到pytracking.parameter.atom.default.py文件的最后,改写deep_feat那句,把net_path改成刚刚移动的那个文件deep_feat = deep.ATOMResNet18(net_path='atom_default.pth', ..原创 2020-08-18 14:46:05 · 1920 阅读 · 2 评论 -
ATOM-resnet50-result
train: bbreg.atom.pydataset: train:LaSOT 全部,val:GOT10k_valnet: atom_resnet50result: 保存在 result_modle[train: 50, 1 / 1000] FPS: 4.5 (4.5) , Loss/total: 0.02596 , Loss/iou: 0.02596[train: 50, 2 / 1000] FPS: 8.7 (105.9) , Loss/total: 0.02344 , .原创 2020-08-08 16:46:11 · 306 阅读 · 0 评论 -
pytracking-ATOM代码训练run_training.py记录
train:bbreg.atom.pydataset: train:LaSOT 全部val:GoT10k_val[train: 50, 1 / 1000] FPS: 3.5 (3.5) , Loss/total: 0.02823 , Loss/iou: 0.02823[train: 50, 2 / 1000] FPS: 6.8 (92.1) , Loss/total: 0.02555 , Loss/iou: 0.02555[train: 50, 3 / 1000] FPS: 9.原创 2020-08-04 20:34:13 · 1639 阅读 · 9 评论 -
pytracking-ATOM代码---OTB100数据集测试结果--GTX1080
(1)run_tracker.py每次可以选择一个类别运行跟踪器,debug=0:保存跟踪结果,debug>0:对跟踪结果进行可视化,测试Bolt2,David如下为debug=5时的最后结果:Bolt2:感觉loss曲线有点奇怪David:(2)run_experiment.pyOTB100全部数据,但其中两个类别,BlurCar1和KiteSurf有问题,注释掉了,所以是在98个序列上运行。其中debug=0保存结果并运行otb_plot.py(根据说明自己创建的),计算AUC分数,原创 2020-08-03 15:38:34 · 2634 阅读 · 14 评论