TasksSotADatasetsPapersMethodsSubmitAbout
Papers With Code 2

A community resource for machine learning research: papers, code, benchmarks, and state-of-the-art results.

Explore

Notable BenchmarksAll SotADatasetsPapersMethods

Community

Submit ResultsAbout

Data sourced from the PWC Archive (CC-BY-SA 4.0). Built by the community, for the community.

Papers/PVT v2: Improved Baselines with Pyramid Vision Transformer

PVT v2: Improved Baselines with Pyramid Vision Transformer

Wenhai Wang, Enze Xie, Xiang Li, Deng-Ping Fan, Kaitao Song, Ding Liang, Tong Lu, Ping Luo, Ling Shao

2021-06-25Panoptic SegmentationImage ClassificationObject Detection
PaperPDFCodeCodeCodeCodeCodeCodeCodeCodeCodeCodeCodeCodeCodeCodeCode(official)CodeCodeCode

Abstract

Transformer recently has presented encouraging progress in computer vision. In this work, we present new baselines by improving the original Pyramid Vision Transformer (PVT v1) by adding three designs, including (1) linear complexity attention layer, (2) overlapping patch embedding, and (3) convolutional feed-forward network. With these modifications, PVT v2 reduces the computational complexity of PVT v1 to linear and achieves significant improvements on fundamental vision tasks such as classification, detection, and segmentation. Notably, the proposed PVT v2 achieves comparable or better performances than recent works such as Swin Transformer. We hope this work will facilitate state-of-the-art Transformer researches in computer vision. Code is available at https://github.com/whai362/PVT.

Results

TaskDatasetMetricValueModel
Object DetectionCOCO-OAverage mAP28.2PVTv2-B5 (Mask R-CNN)
Object DetectionCOCO-OEffective Robustness6.85PVTv2-B5 (Mask R-CNN)
Object DetectionCOCO minivalAP5069.5Sparse R-CNN (PVTv2-B2)
Object DetectionCOCO minivalAP7554.9Sparse R-CNN (PVTv2-B2)
Object DetectionCOCO minivalbox AP50.1Sparse R-CNN (PVTv2-B2)
Image ClassificationImageNetGFLOPs11.8PVTv2-B4
Image ClassificationImageNetGFLOPs6.9PVTv2-B3
Image ClassificationImageNetGFLOPs4PVTv2-B2
Image ClassificationImageNetGFLOPs2.1PVTv2-B1
Image ClassificationImageNetGFLOPs0.6PVTv2-B0
3DCOCO-OAverage mAP28.2PVTv2-B5 (Mask R-CNN)
3DCOCO-OEffective Robustness6.85PVTv2-B5 (Mask R-CNN)
3DCOCO minivalAP5069.5Sparse R-CNN (PVTv2-B2)
3DCOCO minivalAP7554.9Sparse R-CNN (PVTv2-B2)
3DCOCO minivalbox AP50.1Sparse R-CNN (PVTv2-B2)
2D ClassificationCOCO-OAverage mAP28.2PVTv2-B5 (Mask R-CNN)
2D ClassificationCOCO-OEffective Robustness6.85PVTv2-B5 (Mask R-CNN)
2D ClassificationCOCO minivalAP5069.5Sparse R-CNN (PVTv2-B2)
2D ClassificationCOCO minivalAP7554.9Sparse R-CNN (PVTv2-B2)
2D ClassificationCOCO minivalbox AP50.1Sparse R-CNN (PVTv2-B2)
2D Object DetectionCOCO-OAverage mAP28.2PVTv2-B5 (Mask R-CNN)
2D Object DetectionCOCO-OEffective Robustness6.85PVTv2-B5 (Mask R-CNN)
2D Object DetectionCOCO minivalAP5069.5Sparse R-CNN (PVTv2-B2)
2D Object DetectionCOCO minivalAP7554.9Sparse R-CNN (PVTv2-B2)
2D Object DetectionCOCO minivalbox AP50.1Sparse R-CNN (PVTv2-B2)
16kCOCO-OAverage mAP28.2PVTv2-B5 (Mask R-CNN)
16kCOCO-OEffective Robustness6.85PVTv2-B5 (Mask R-CNN)
16kCOCO minivalAP5069.5Sparse R-CNN (PVTv2-B2)
16kCOCO minivalAP7554.9Sparse R-CNN (PVTv2-B2)
16kCOCO minivalbox AP50.1Sparse R-CNN (PVTv2-B2)

Related Papers

Automatic Classification and Segmentation of Tunnel Cracks Based on Deep Learning and Visual Explanations2025-07-18Adversarial attacks to image classification systems using evolutionary algorithms2025-07-17Efficient Adaptation of Pre-trained Vision Transformer underpinned by Approximately Orthogonal Fine-Tuning Strategy2025-07-17Federated Learning for Commercial Image Sources2025-07-17MUPAX: Multidimensional Problem Agnostic eXplainable AI2025-07-17A Real-Time System for Egocentric Hand-Object Interaction Detection in Industrial Domains2025-07-17RS-TinyNet: Stage-wise Feature Fusion Network for Detecting Tiny Objects in Remote Sensing Images2025-07-17Decoupled PROB: Decoupled Query Initialization Tasks and Objectness-Class Learning for Open World Object Detection2025-07-17