Insights from Autoresearch for Solar Panel Segmentation

Justinas Lekavicius; Kursat Komurcu; Valentas Gruzauskas; Linas Petkevicius

cs.CVOct 7, 2026

Insights from Autoresearch for Solar Panel Segmentation

Authors: Justinas Lekavicius, Kursat Komurcu, Valentas Gruzauskas, Linas Petkevicius

Organizations: Vilnius University, Institute of Computer Science, Artificial Intelligence Methods Lab · IRISA, Université Bretagne Sud · European Commission Joint Research Centre

Abstract

This paper investigates AutoResearch, a protocol in which a coding language model edits a training program under a one-hour GPU budget and retains a change only if validation IoU improves. The protocol is applied to photovoltaic panel segmentation on a frozen real-image split, with DeepLabV3--ResNet-50 held fixed. Three campaigns of 24 experiments, using Gemma~4 12B, Qwen3-8B all improve their one-hour baselines, but retained modifications do not transfer across hardware. The Qwen3-8B configuration, trained on real images only, reaches a test IoU of 0.836 versus 0.833 for the reference GAN-augmented schedule. Research repository https://github.com/VU-AIML/automl4eo-autoresearch-segmentation.

Figures & tables

	Loop A	Loop B	Loop C
Coding model	Gemma 4 12B	Qwen3-8B	Qwen3-8B
GPU	RTX 4070 12 GB	ASUS Ascent GX10	ASUS Ascent GX10
Batch / peak VRAM	2 / $\approx$ 9 GB	8 / 18–35 GB	8 $\to$ 16 / 35–36 GB
Baseline val. IoU	0.616	0.723	0.737
Best val. IoU	0.710	0.858	0.858
$\Delta$ val. IoU	$+0.094$	$+0.135$	$+0.120$

Table 1: Summary of the three AutoResearch campaigns. Architecture, split, seed, and a one-hour training budget are shared. Loops A and C are not evaluated on the held-out test set.

Figure 1: Validation IoU over three AutoResearch campaigns on the same frozen real split and one-hour budget. Solid lines show the running best of each campaign. Transparent markers correspond to rejected trials. Stars mark the selected best of each campaign. The Loop A clipping trial (IoU 0.349) lies below the plotted range.

Modification family	Loop A	Loop B	Loop C
Soft Dice (+BCE)	rejected	retained $+0.103$	retained $+0.076$
Cosine schedule	retained $+0.092$	rejected	rejected
Polynomial LR 0.9	rejected	retained $+0.012$	—
AdamW vs. Adam	rejected	retained $+0.005$	—
Auxiliary head $\times 0.4$	rejected	retained $+0.004$	—
Gradient clip 1.0	rejected $-0.359$	retained $+0.002$	—

Table 2: Outcome of overlapping modification families. Deltas are computed against the configuration on which the change was stacked. Retention means that the commit became the new incumbent.

Setup	Data	Budget	IoU	F1
Paper no_aug	real	$\leq$ 100 ep. (A100)	0.801	0.853
Paper basic_aug	real	$\leq$ 100 ep. (A100)	0.813	0.865
Paper gan60 (paper best)	real + 60% GAN	$\leq$ 100 ep. (A100)	0.833	0.880
Loop C best (val.)	real	3600 s (GX10)	0.858 †	—
Loop B best (val.)	real	3600 s (GX10)	0.858 †	0.908 †
Loop B best (test)	real	3600 s (GX10)	0.836	0.891

Table 3: Loop B compared with Table 5 of Lekavičius and Gružauskas (2024) on the same 256 test images. Loops A and C are validation-only.

Explore similar work

Vision Transformer Ensembles for Panoramic Street Segmentation
Oct 5, 2026Yunus Serhat BıçakçıPanoptic Segmentation
Department of Artificial Intelligence and Machine Learning, Faculty of Applied Sciences, Marmara University, Istanbul, Türkiye · Geospatial Data Science Group, School of Geographical and Earth Sciences, University of Glasgow, Glasgow, UK
LlamaSeg: Image Segmentation via Autoregressive Mask Generation
May 26, 2025Jiru Deng, Tengjin Weng, Tianyu Yang +3Visual Autoregressive Models Autoregressive Generation
Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ) · Tsinghua University · Shenzhen University +2
EOVSAM: Efficient Open-Vocabulary Segmentation with SAM 3 in One Pass
Aug 3, 2026Haomin Peng, Yongkang Li, Zhaoxiang Liu +4Open-Vocabulary Segmentation Open-Vocabulary
Huazhong University of Science and Technology · Data Science & Artificial Intelligence Research Institute, China Unicom · Unicom Data Intelligence, China Unicom +1

Oct 5, 2026, cs.CVJ/K move · Enter open · S save

Vision Transformer Ensembles for Panoramic Street Segmentation

Yunus Serhat Bıçakçı

Department of Artificial Intelligence and Machine Learning, Faculty of Applied Sciences, Marmara University, Istanbul, Türkiye · Geospatial Data Science Group, School of Geographical and Earth Sciences, University of Glasgow, Glasgow, UK

Panoptic Segmentation

Open paper AskPDF arXiv

Semantic segmentation of street panoramas can support detailed descriptions of urban environments, yet small datasets and unequal training costs make model selection difficult. This paper presents the system used for a first place submission to the PalmCity challenge in the leaderboard snapshot dated 5 October 2026. Nine pretrained segmentation systems are compared using approximately equal computation budgets. The candidates include DeepLabV3+, SegFormer, UPerNet, Mask2Former, DINOv3 with a linear decoder, and an Encoder only Mask Transformer using DINOv3. The two leading candidates are trained independently with three random seeds and longer budgets. Equal averaging of class probabilities from the three Encoder only Mask Transformer models, evaluated at three image scales with horizontal reflection, produces 60.95% mean intersection over union and 71.16% mean F1 on the 84 image public validation split. The submitted predictions receive 57.08% mean intersection over union and 67.96% mean F1 on the hidden test leaderboard. Producing all 249 test masks takes 251.49 seconds including model initialization and provenance checks on one NVIDIA RTX 5090. Peak allocated GPU memory is 2.70 GiB. The study reports all eligible models, all inference variants, class level errors, source conditions, and reproducibility checks, providing a documented challenge workflow with existing architectures.

Abstract

Explore similar work

Vision Transformer Ensembles for Panoramic Street Segmentation

LlamaSeg: Image Segmentation via Autoregressive Mask Generation

EOVSAM: Efficient Open-Vocabulary Segmentation with SAM 3 in One Pass