hahattpro / crowd_counting_framework (public) (License: Unspecified) (since 2020-12-23) (hash sha1)
backupfrom https://github.com/ttpro1995/crowd_counting_framework
Clone URLs: https://rocketgit.com/user/hahattpro/crowd_counting_framework ssh://rocketgit@ssh.rocketgit.com/user/hahattpro/crowd_counting_framework git://git.rocketgit.com/user/hahattpro/crowd_counting_framework
debug/a-1 develop feature/add_can_adcrowdnet feature/add_context_aware_network feature/attn_can_adcrowd_simple feature/cache_dataloader feature/compact_cnn feature/compare_with_original_count feature/dataloader_not_divide_8 feature/debug_viz feature/documentation feature/improve_density_gen feature/improve_loss_fn feature/loader_full_and_augment feature/refactor-dataflow feature/refactor_ccnn_basecode feature/save_best feature/simple_cometml feature/train_context_aware_network_on_shtA feature/train_dev_test_split hotfix/fix_visualize_data_loader master meow_exp server-training with_apex
List of commits:
Subject Hash Author Date (UTC)
change optim to Adam, lr and decay same with content aware network 0b0879b22e9d202015e67a0cc8c87ddcafb3c2d2 Thai Thien 2020-02-28 17:15:05
train with higher learning rate 3fa7e328dc4b7b4570fd32908b3df5fab8283919 Thai Thien 2020-02-28 16:57:16
train 25 epoch only 1cfb58b5eae299f49e3b64289c3a12a8f35cee45 Thai Thien 2020-02-27 18:35:58
we freeze vgg pretrain layer 459cefbfbf53949cb9636ca002cd4f78e964bb83 Thai Thien 2020-02-27 18:24:17
fix error when save 86a8f467d47a67da0accc963e8baacbaea07ab14 Thai Thien 2020-02-27 17:55:46
we discard deform 7 and keep deform 5 instead 91ff78e4a1aef3677b15447b631c2927c9e62a31 Thai Thien 2020-02-27 16:22:47
reduce image by crop 530516ff40b744cac79756c97faaed50c8b65cfb Thai Thien 2020-02-27 16:19:11
try to solve x8/8 dimension mismatch 6b0616197995b3c95041cde43030245f96f1f81e Thai Thien 2020-02-27 16:17:12
reduce model size, discard middle deform (kernal size 5) 00fb477cf699a8f2eed8e861f241c077b2046000 Thai Thien 2020-02-27 16:11:51
shanghaitech same size density map 25fc2e9395dd67bb6cc273c7fe6cd64c3b154120 Thai Thien 2020-02-27 16:03:05
fix target1.unsqueeze(0) bb5c46f121207fd9838ee085314df40513e2ed32 Thai Thien 2020-02-27 15:56:07
because GPU too small, we crop shanghaitech A 0f97517dd8de5d3d75cf49af9aa9d20e4595ea3e Thai Thien 2020-02-27 15:52:21
meow 3b9576511e597b4894c7c95ca06dc5219ef9c524 tthien 2020-02-27 15:44:39
fix wrong py name in script, dataset use only 1 worker 3bd765c9aa79e47049a94547e57deb68c947001c Thai Thien 2020-02-27 15:41:44
minor fix 5505b9a8b28df105e11a99d3462323ae9d507102 Thai Thien 2020-02-27 15:28:37
get ready for short training run with 30 epochs 66dda0858561897cd5f81e10077459adb39d86dd Thai Thien 2020-02-27 15:22:01
implement attn_can_adcrowdnet ffd38664a43d861c20cdc225746b9ce2a00260c7 Thai Thien 2020-02-27 15:10:27
WIP: add can-adcrowdnet 5620b83449b31d00a367c8de77e431e19a5ccfb3 Thai Thien 2020-02-25 11:31:38
add readable timestamp viz ae1fdb49ddb9ea77659529dceb7fb87c2790c8dc Thai Thien 2020-02-24 03:49:57
change save name prefix c53a86f30fb8fd4e8f3a409eb67827d56a43ae5c Thai Thien 2020-02-02 10:48:15
Commit 0b0879b22e9d202015e67a0cc8c87ddcafb3c2d2 - change optim to Adam, lr and decay same with content aware network
Author: Thai Thien
Author date (UTC): 2020-02-28 17:15
Committer name: Thai Thien
Committer date (UTC): 2020-02-28 17:15
Parent(s): 3fa7e328dc4b7b4570fd32908b3df5fab8283919
Signing key:
Tree: f25c4edf02c873718fbcb643e57fa85ec195b950
File Lines added Lines deleted
train_attn_can_adcrowdnet_freeze_vgg.py 1 2
train_script/attn_can_adcrowdnet/train_server_31_epoch_shA_freeze_vgg.sh 1 0
File train_attn_can_adcrowdnet_freeze_vgg.py changed (mode: 100644) (index de7e0fb..e47bfc7)
... ... if __name__ == "__main__":
38 38 # loss function # loss function
39 39 loss_fn = nn.MSELoss(size_average=False).to(device) loss_fn = nn.MSELoss(size_average=False).to(device)
40 40
41 optimizer = torch.optim.SGD(model.parameters(), args.lr,
42 momentum=args.momentum,
41 optimizer = torch.optim.Adam(model.parameters(), args.lr,
43 42 weight_decay=args.decay) weight_decay=args.decay)
44 43
45 44 trainer = create_supervised_trainer(model, optimizer, loss_fn, device=device) trainer = create_supervised_trainer(model, optimizer, loss_fn, device=device)
File train_script/attn_can_adcrowdnet/train_server_31_epoch_shA_freeze_vgg.sh changed (mode: 100644) (index 5ceb244..cb7dac6)
... ... CUDA_VISIBLE_DEVICES=5 nohup python train_attn_can_adcrowdnet_freeze_vgg.py \
10 10 --task_id attn_can_adcrowdnet_default_shtA_33_freeze_vgg_2 \ --task_id attn_can_adcrowdnet_default_shtA_33_freeze_vgg_2 \
11 11 --input /data/rnd/thient/thient_data/ShanghaiTech/part_A \ --input /data/rnd/thient/thient_data/ShanghaiTech/part_A \
12 12 --lr 1e-4 \ --lr 1e-4 \
13 --decay 5*1e-4 \
13 14 --datasetname shanghaitech \ --datasetname shanghaitech \
14 15 --epochs 33 > logs/attn_can_adcrowdnet_default_shtA_33_freeze_vgg_2.log & --epochs 33 > logs/attn_can_adcrowdnet_default_shtA_33_freeze_vgg_2.log &
Hints:
Before first commit, do not forget to setup your git environment:
git config --global user.name "your_name_here"
git config --global user.email "your@email_here"

Clone this repository using HTTP(S):
git clone https://rocketgit.com/user/hahattpro/crowd_counting_framework

Clone this repository using ssh (do not forget to upload a key first):
git clone ssh://rocketgit@ssh.rocketgit.com/user/hahattpro/crowd_counting_framework

Clone this repository using git:
git clone git://git.rocketgit.com/user/hahattpro/crowd_counting_framework

You are allowed to anonymously push to this repository.
This means that your pushed commits will automatically be transformed into a merge request:
... clone the repository ...
... make some changes and some commits ...
git push origin main