AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

GiT: Towards Generalist Vision Transformer through Universal Language Interface

Haiyang Wang, Hao Tang, Li Jiang, Shaoshuai Shi, Muhammad Ferjad Naeem, Hongsheng Li, Bernt Schiele, Liwei Wang

ECCV 2024 · Vision-language and multimodal › CLIP and zero-shot · approach: Adapting VLMs for zero-shot classification · 12 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
joint multi-task training of five tasks instead of single-task trainingGiT-B · COCO2017 detectionAP: 45.1 -> 46.7Improved
joint multi-task training instead of single-task trainingGiT-B · ADE20K semantic segmentationmIoU: 47.7 -> 47.8No clear effect
joint multi-task training instead of single-task trainingGiT-B · COCO CaptionCIDEr: 107.9 -> 112.6Improved
scale model from ViT-B to ViT-HGiT multi-task · COCO2017 detectionAP: 46.7 -> 52.9Improved
universal training on 27 datasets instead of five-dataset multi-task trainingGiT-B · Cityscapes semantic segmentation (zero-shot)mIoU: 34.4 -> 56.2Improved
universal training instead of multi-task trainingGiT-B · DRIVE few-shot (K=5)mDice: 34.3 -> 51.1Improved
use decoder-only architecture instead of encoder-decoder (ablation)GiT-B multi-task · ADE20K semantic segmentationmIoU: 46.9 -> 47.8Improved
add modality experts (multiple FFN) (ablation)GiT-B multi-task · COCO2017 detectionAP: 46.1 -> 46.2No clear effect
add text conditioning (image-to-text attention) (ablation)GiT-B multi-task · RefCOCO visual groundingAcc@0.5: 78.7 -> 85.8Improved
add new transformer layers on top of SAM-initialized ViT (ablation)GiT-B single-task · COCO2017 detectionAP: 40.2 -> 45.1Improved
use accelerated global attention over shared observations only (ablation)GiT-B single-task · ADE20K semantic segmentationmIoU and training time: 47.9 -> 47.7; 51h -> 35hMixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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