Sovereign 8s.is mirror of standardgalactic/Class-Incremental-Few-Shot-Underwater-Object-Detection-Framework
Find a file
Repository files (latest commit first)
Filename Latest commit message Latest commit date
2026-02-19 15:52:38 +08:00
fs Add files via upload 2026-02-19 04:57:28 +08:00
utils Add files via upload 2026-02-19 04:57:28 +08:00
README.md Fix command path in README for evaluation 2026-02-19 15:52:38 +08:00
test_fasterrcnn.py Add files via upload 2026-02-19 04:57:28 +08:00

Class-Incremental-Few-Shot-Underwater-Object-Detection-Framework

To this end, this paper proposes a class-incremental few-shot underwater object detection framework(CIFS-UD) This repository provides the official implementation of CIFS-UD, a Class-Incremental Few-Shot Underwater Object Detection framework designed to address catastrophic forgetting and data scarcity in underwater object detection.

CIFS-UD

Class-Incremental Few-Shot Underwater Object Detection

Python PyTorch CUDA License


📌 Overview

CIFS-UD is a Class-Incremental Few-Shot Underwater Object Detection framework designed to address:

  • Catastrophic forgetting in class-incremental learning
  • Data scarcity in underwater environments
  • Prototype-guided incremental adaptation

The framework adopts a three-stage training paradigm:

  1. Base Training
  2. Prototype Extraction
  3. Incremental Few-Shot Fine-Tuning

🗂 Dataset Preparation

Supported datasets:

  • Brackish
  • TrashCan
  • RUOD

All datasets must be converted to Pascal VOC 2007 format.

Directory structure:


⚙️ Installation

Requirements

  • Python >= 3.8
  • CUDA 11.8
  • PyTorch 2.0.0

Install PyTorch:

pip install torch==2.0.0+cu118 torchvision==0.15.1+cu118 --index-url https://download.pytorch.org/whl/cu118

Evaluation

python test_fasterrcnn.py --config-file configs/RUOD/split1/inc/10shot_INC_CLP.yml --eval-only