Iterative Filter Pruning for Concatenation-based CNN Architectures

Svetlana Pavlitska, Oliver Bagge, Federico Peccia, Toghrul Mammadov, J. Marius Zöllner
5/4/2024
cs.CV

Official code: fzi-forschungszentrum-informatik/iterative-yolo-pruning (8 stars, GPL-3.0, last commit 2024-05-13).

Abstract

Model compression and hardware acceleration are essential for the resource-efficient deployment of deep neural networks. Modern object detectors have highly interconnected convolutional layers with concatenations. In this work, we study how pruning can be applied to such architectures, exemplary for YOLOv7. We propose a method to handle concatenation layers, based on the connectivity graph of convolutional layers. By automating iterative sensitivity analysis, pruning, and subsequent model fine-tuning, we can significantly reduce model size both in terms of the number of parameters and FLOPs, while keeping comparable model accuracy. Finally, we deploy pruned models to FPGA and NVIDIA Jetson Xavier AGX. Pruned models demonstrate a 2x speedup for the convolutional layers in comparison to the unpruned counterparts and reach real-time capability with 14 FPS on FPGA. Our code is available at https://github.com/fzi-forschungszentrum-informatik/iterative-yolo-pruning.

View on arXivView PDF

Code Implementations(1)

fzi-forschungszentrum-informatik/iterative-yolo-pruningOfficial100%
GPL-3.0

Iterative Filter Pruning for Concatenation-Based CNN Architectures

81Python, Jupyter NotebookApr 16, 20242 years agoGPL-3.0

Similar Papers

Explainability-Inspired Layer-Wise Pruning of Deep Neural Networks for Efficient Object Detection

85% match2026

REPrune: Channel Pruning via Kernel Representative Selection

84% match2024

LayerMerge: Neural Network Depth Compression through Layer Pruning and Merging

84% match1 repo322024

Compressing CNN models for resource-constrained systems by channel and layer pruning

83% match2025

Pick-or-Mix: Dynamic Channel Sampling for ConvNets

83% match1 repo132026

RL-Pruner: Structured Pruning Using Reinforcement Learning for CNN Compression and Acceleration

83% match1 repo292024

Cite this paper

@article{pavlitska2024iterative,
  title  = {Iterative Filter Pruning for Concatenation-based CNN Architectures},
  author = {Svetlana Pavlitska and Oliver Bagge and Federico Peccia and Toghrul Mammadov and J. Marius Zöllner},
  year   = {2024},
  eprint = {2405.03715},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2405.03715v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2405.03715.svg)](https://codeofpaper.com/paper/2405.03715)

Discussion