Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

edit #3

Open
wants to merge 5 commits into
base: main
Choose a base branch
from
Open

edit #3

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
69da2e4
Update 2) Gesture Translation Lit Review
ImaadM097 Sep 13, 2022
044621e
Update 2) Gesture Translation Lit Review
ImaadM097 Sep 13, 2022
b5795ce
Update 2) Gesture Translation Lit Review
ImaadM097 Sep 13, 2022
514e6b0
Merge branch 'ASCII-Mentorships:main' into main
ImaadM097 Sep 13, 2022
7fdd119
Update 2) Gesture Translation Lit Review
ImaadM097 Sep 15, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 17 additions & 2 deletions 2) Gesture Translation Lit Review
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,5 +6,20 @@
Architecture: Faster R-CNN + 3D Conv + LSTM. (Not very feasible for us)
Accuracy: 99% on common vocabulary data set.
Link: https://ieeexplore.ieee.org/document/8950864


3. Title:- Neural Sign Language Translation (nslt)
Dataset used:- Continuous SLT dataset, RWTHPHOENIX-Weather 2014T 1 (A set of photos and videos that provide translations of german sign language weather forecasts, around 30 GB in size). Link-
https://www-i6.informatik.rwth-aachen.de/~koller/RWTH-PHOENIX-2014-T/
Accuracy:- Accuracy has been mentioned in the terms of BLEU-4. The upper bound for translation performance is 19.26 BLEU-4
Code:- Code is available in the GitHub repo:- https://github.com/neccam/nslt
Implementation:- The experiments were conducted by grouping them in three categories.
1. Gloss2Text (G2T), in which we simulate having a perfect SLR system as an intermediate tokenization.
2. Sign2Text (S2T) which covers the end-to-end pipeline translating directly from frame-level sign language video into spoken language.
3. Sign2Gloss2Text (S2G2T) which uses a SLR system as tokenization layer to add intermediate supervision.
Link:- https://openaccess.thecvf.com/content_cvpr_2018/papers/Camgoz_Neural_Sign_Language_CVPR_2018_paper.pdf
4. Title:- Word-level Deep Sign Language Recognition from Video: A New Large-scale Dataset and Methods Comparison
Dataset used:- Word-Level American Sign Language (WLASL) (Collection videos of more than 2000 words by different signers) https://dxli94.github.io/WLASL/
Consists of only RGB videos. Around 34,404 video samples of 3,126 glosses for further annotations.
Implementation:- The model has been made by using temporal graph convolutional network (TGCN). The models, VGG-GRU, Pose-GRU, Pose-TGCN and
I3D are implemented in PyTorch. The ratio of training, validation and testing set was 4:1:1.
Accuracy:- 62.63% at top-10 accuracy on 2,000words/glosses
Link:- with code- https://github.com/dxli94/WLASL