Explicit Visual Prompting for Universal Foreground Segmentations

29 May 2023  ·  Weihuang Liu, Xi Shen, Chi-Man Pun, Xiaodong Cun ·

Foreground segmentation is a fundamental problem in computer vision, which includes salient object detection, forgery detection, defocus blur detection, shadow detection, and camouflage object detection. Previous works have typically relied on domain-specific solutions to address accuracy and robustness issues in those applications. In this paper, we present a unified framework for a number of foreground segmentation tasks without any task-specific designs. We take inspiration from the widely-used pre-training and then prompt tuning protocols in NLP and propose a new visual prompting model, named Explicit Visual Prompting (EVP). Different from the previous visual prompting which is typically a dataset-level implicit embedding, our key insight is to enforce the tunable parameters focusing on the explicit visual content from each individual image, i.e., the features from frozen patch embeddings and high-frequency components. Our method freezes a pre-trained model and then learns task-specific knowledge using a few extra parameters. Despite introducing only a small number of tunable parameters, EVP achieves superior performance than full fine-tuning and other parameter-efficient fine-tuning methods. Experiments in fourteen datasets across five tasks show the proposed method outperforms other task-specific methods while being considerably simple. The proposed method demonstrates the scalability in different architectures, pre-trained weights, and tasks. The code is available at: https://github.com/NiFangBaAGe/Explicit-Visual-Prompt.

PDF Abstract

Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Camouflaged Object Segmentation CAMO EVPv2 MAE 0.058 # 2
Weighted F-Measure 78.6 # 2
S-Measure 84.8 # 1
E-Measure 89.9 # 2
Camouflaged Object Segmentation COD EVPv2 MAE 0.029 # 1
Weighted F-Measure 74.6 # 1
S-Measure 84.3 # 1
E-Measure 90.9 # 1
Salient Object Detection DUT-OMRON EVPv2 max_F1 0.857 # 2
MAE 0.047 # 4
E-measure 0.895 # 1
S-measure 0.862 # 1
Salient Object Detection DUTS-TE EVPv2 MAE 0.027 # 3
max_F1 0.923 # 1
E-measure 0.948 # 1
S-measure 0.915 # 6
Salient Object Detection ECSSD EVPv2 MAE 0.028 # 3
max_F1 0.958 # 2
S-measure 0.935 # 1
E-measure 0.957 # 1
Salient Object Detection HKU-IS EVPv2 MAE 0.023 # 1
E-measure 0.963 # 1
max_F1 0.953 # 1
S-measure 0.932 # 1
Salient Object Detection PASCAL-S EVPv2 MAE 0.053 # 2
max_F1 0.869 # 6
S-measure 0.879 # 1
E-measure 0.917 # 1

Methods


No methods listed for this paper. Add relevant methods here