Exploring Model Quantization in GenAI-based Image Inpainting and Detection of Arable Plants

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Modak, Sourav, Saltık, Ahmet Oğuz, Stein, Anthony
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910857733079040
author Modak, Sourav
Saltık, Ahmet Oğuz
Stein, Anthony
author_facet Modak, Sourav
Saltık, Ahmet Oğuz
Stein, Anthony
contents Deep learning-based weed control systems often suffer from limited training data diversity and constrained on-board computation, impacting their real-world performance. To overcome these challenges, we propose a framework that leverages Stable Diffusion-based inpainting to augment training data progressively in 10% increments -- up to an additional 200%, thus enhancing both the volume and diversity of samples. Our approach is evaluated on two state-of-the-art object detection models, YOLO11(l) and RT-DETR(l), using the mAP50 metric to assess detection performance. We explore quantization strategies (FP16 and INT8) for both the generative inpainting and detection models to strike a balance between inference speed and accuracy. Deployment of the downstream models on the Jetson Orin Nano demonstrates the practical viability of our framework in resource-constrained environments, ultimately improving detection accuracy and computational efficiency in intelligent weed management systems.
format Preprint
id arxiv_https___arxiv_org_abs_2503_02420
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Exploring Model Quantization in GenAI-based Image Inpainting and Detection of Arable Plants
Modak, Sourav
Saltık, Ahmet Oğuz
Stein, Anthony
Computer Vision and Pattern Recognition
Artificial Intelligence
Deep learning-based weed control systems often suffer from limited training data diversity and constrained on-board computation, impacting their real-world performance. To overcome these challenges, we propose a framework that leverages Stable Diffusion-based inpainting to augment training data progressively in 10% increments -- up to an additional 200%, thus enhancing both the volume and diversity of samples. Our approach is evaluated on two state-of-the-art object detection models, YOLO11(l) and RT-DETR(l), using the mAP50 metric to assess detection performance. We explore quantization strategies (FP16 and INT8) for both the generative inpainting and detection models to strike a balance between inference speed and accuracy. Deployment of the downstream models on the Jetson Orin Nano demonstrates the practical viability of our framework in resource-constrained environments, ultimately improving detection accuracy and computational efficiency in intelligent weed management systems.
title Exploring Model Quantization in GenAI-based Image Inpainting and Detection of Arable Plants
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2503.02420