Msmsfnet: a multi-stream and multi-scale fusion net for edge detection

2024-04-07

arXiv.org

AI Summary

Existing edge detection methods heavily rely on ImageNet pre-trained weights, limiting architectural flexibility and failing on non-natural imagery such as synthetic aperture radar (SAR) data. To address this, we propose the first end-to-end edge detection framework trained from scratch—without ImageNet pretraining—featuring a multi-stream, multi-scale feature extractor with pyramid fusion and a lightweight decoder under full supervision. We systematically validate, for the first time, the feasibility and superiority of the no-pretraining paradigm on BSDS500, NYUDv2, and Multicue, achieving state-of-the-art performance. On SAR imagery, our method significantly outperforms transfer-learning baselines. Even when compared to ImageNet-pretrained counterparts, it remains competitive on BSDS500. This work establishes a reproducible, low-dependency paradigm for cross-domain edge detection—particularly under heterogeneous optical/SAR distributions—opening new avenues for domain-agnostic model design.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Learning on the Edge & Model Compression
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Systems and Infrastructure for Web, Mobile and WoT: Cloud, edge and content delivery systems for the Web

Abstract

Edge detection is a long-standing problem in computer vision. Despite the efficiency of existing algorithms, their performance, however, rely heavily on the pre-trained weights of the backbone network on the ImageNet dataset. The use of pre-trained weights in previous methods significantly increases the difficulty to design new models for edge detection without relying on existing well-trained ImageNet models, as pre-training the model on the ImageNet dataset is expensive and becomes compulsory to ensure the fairness of comparison. Besides, the pre-training and fine-tuning strategy is not always useful and sometimes even inaccessible. For instance, the pre-trained weights on the ImageNet dataset are unlikely to be helpful for edge detection in Synthetic Aperture Radar (SAR) images due to strong differences in the statistics between optical images and SAR images. Moreover, no dataset has comparable size to the ImageNet dataset for SAR image processing. In this work, we study the performance achievable by state-of-the-art deep learning based edge detectors in publicly available datasets when they are trained from scratch, and devise a new network architecture, the multi-stream and multi-scale fusion net (msmsfnet), for edge detection. We show in our experiments that by training all models from scratch, our model outperforms state-of-the-art edge detectors in three publicly available datasets. We also demonstrate the efficiency of our model for edge detection in SAR images, where no useful pre-trained weight is available. Finally, We show that our model is able to achieve competitive performance on the BSDS500 dataset when the pre-trained weights are used.

Problem

• Edge detection without relying on pre-trained ImageNet models
• Improving edge detection in SAR images lacking pre-trained weights
• Designing a new network architecture for superior edge detection

Innovation

• Multi-stream and multi-scale fusion network
• Training from scratch without pre-trained weights
• Effective for SAR and optical image edge detection