tfzhou/ContrastiveSeg
ICCV2021 (Oral) - Exploring Cross-Image Pixel Contrast for Semantic Segmentation
PythonMIT
Issues
- 6
Memory and semi-hard example mining code
#31 opened by TheoPis - 1
About the dimension of feature embedding
#64 opened by chen-wolfberry - 5
maybe bug in ./segmentor/trainer_contrastive.py
#55 opened by eraserNut - 4
- 0
CKPT
#69 opened by geek-APTX4869 - 4
@tfzhou So, the class 0 is the ignore index and not the background class? I mean, for the background class is applied also the contrastive loss. Right?
#66 opened by zhangyuxuan321 - 7
T-SNE of features visualization
#42 opened by HuadongTang - 2
Example for max_views, max_samples?
#61 opened by lennart-maack - 0
Dataloader distributed or not
#67 opened by zhd2rng - 4
About args n_view,max_samples
#33 opened by xianxuan-z - 3
- 1
- 3
Semi-Hard Example Sampling implementation
#59 opened by nysp78 - 4
code confusing me in def._sample_negative
#60 opened by ck6698000 - 0
- 0
- 3
Questions about T-SNE visualization
#36 opened by dongdongtong - 3
About the hard anchor samping
#43 opened by leeh43 - 7
an AttributeError occurred suddently during train with cityscapes dataset...
#41 opened by shanhuhaifeng - 2
Momentum Update
#57 opened by Macc520 - 1
How to reproduce the SoTA result of Cityscapes?
#58 opened by JasiRose - 2
loss_contrast
#54 opened by Summer77723 - 1
关于特征可视化的代码
#50 opened by NJNUCS - 1
T-SNE of features visualization
#53 opened by xiewende - 3
about using your model in other dataset.
#48 opened by Lemonweier - 0
- 2
The L2 normalization of features
#52 opened by ccccly - 2
- 0
Not able to run train file. Will please share the tips to run coco script step wise
#49 opened by HamzaFarooq013 - 8
about the decrease of the loss
#44 opened by hhxzzy - 0
about the memory bank
#47 opened by hhxzzy - 0
- 1
- 1
can this code run on a lower torch version?
#40 opened by hhxzzy - 6
- 2
network_stride & _dequeue_and_enqueue
#28 opened by Darcy103 - 3
when training code will be available?
#23 opened by seyeeet - 1
CUDA out of memory
#32 opened by xiewende - 1
About simple reproduction
#34 opened by xianxuan-z - 3
why ignore zero?
#35 opened by Junshan233 - 1
Cannot run the loss of loss_contrast_mem.py
#37 opened by winycg - 1
tSNE visualisation
#26 opened by YuemingJin - 3
Explanation of n_view
#24 opened by HenryPengZou - 1
- 1
- 1
- 1
- 3
Question about hard sample mining
#22 opened by curiosity654 - 5
Question about the subtraction of max from inner product in the loss computation
#20 opened by edwardpwtsoi - 0
Explanation of n_view and Specific Dimension of the input of Contrastive function
#21 opened by HenryPengZou