AI Paper Atlas

Research map / Video / Video segmentation and tracking

Visual Transformation Aided Contrastive Learning for Video-Based Kinship Verification

Hamdi Dibeklioglu

ICCV 2017 · Video › Video segmentation and tracking · approach: Deepfake video detection · 30 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add expression matching of sub-sequences (vs random five-frame pairs) (ablation)proposed method · UvA-NEMO Smile (video kinship)accuracy (%): 82.74 -> 93.65Improved
use temporal sequences of 5 frames instead of single frames (m=0) (ablation)proposed method · UvA-NEMO Smile (video kinship)accuracy (%): 89.51 -> 93.65Improved
increase sequence length from 5 to 7 frames (ablation)proposed method · UvA-NEMO Smile (video kinship)accuracy (%): 93.65 -> 91.21Hurt
increase sequence length from 1 to 3 frames (ablation)proposed method · UvA-NEMO Smile (video kinship)accuracy (%): 89.51 -> 92.93Improved
add non-kinship loss with 10% most similar non-kins (ablation)proposed method · UvA-NEMO Smile (video kinship)accuracy (%): 90.28 -> 93.65Improved
use 10% instead of 100% of non-kins in non-kinship loss (ablation)proposed method · UvA-NEMO Smile (video kinship)accuracy (%): 91.31 -> 93.65Improved
use 10% instead of 50% of non-kins in non-kinship loss (ablation)proposed method · UvA-NEMO Smile (video kinship)accuracy (%): 92.15 -> 93.65Improved
add contrastive loss in both pre-training and fine-tuning (ablation)proposed method · UvA-NEMO Smile (video kinship)accuracy (%): 86.30 -> 93.65Improved
use appearance-based contrastive pre-training instead of standard (ablation)proposed method · UvA-NEMO Smile (video kinship)accuracy (%): 90.28 -> 93.65Improved
use feature-based contrastive fine-tuning instead of standard (logistic regression) (ablation)proposed method · UvA-NEMO Smile (video kinship)accuracy (%): 92.72 -> 93.65Improved
replace coupled encoder-decoder with stacked denoising autoencoders (ablation)proposed method · UvA-NEMO Smile (video kinship)accuracy (%): 93.65 -> 86.40Hurt
use short-term temporal video information vs static images (SMCNN) (ablation)SMCNN baseline · UvA-NEMO Smile (video kinship)accuracy (%): 84.56 -> 93.65Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy