Melola Discovery

← Melola Discovery

Akses terbuka Jurnal SINTA 1

Decoder-only transformer with multi-scale attention for efficient printed circuit board defect inspection

Bulletin of Electrical Engineering and Informatics · Vol. 15 No. 4 (2026) · hlm. 3070–3079

DOI
10.11591/eei.v15i4.10873
ISSN
2089-3191, 2302-9285
Terbit
2026-08-01
Penerbit
Institute of Advanced Engineering and Science (IAES)
Bahasa
en
Lisensi
creativecommons.org/licenses/by-sa/4.0/

Abstrak

Printed circuit boards (PCBs) are critical components in modern electronics, detecting defects rapidly, and accurately is crucial in manufacturing. We propose a novel real-time PCB defect detection model based on a Transformer decoder-only architecture. The framework first extracts multi-scale features via a backbone with a feature pyramid network (FPN) and generates candidate regions using a region proposal network (RPN). These region proposals are encoded as query embeddings that drive an adaptive multi-scale deformable attention (AMDA) module in the Transformer decoder, replacing the standard encoder-decoder attention. By dynamically weighting multi-scale feature maps for each query, AMDA emphasizes the feature scale most relevant to each defect’s size and texture, and yielding enhanced discriminative representations for subtle defect detection. The decoder-only design drastically reduces computational overhead compared to full encoder-decoder Transformers, enabling real-time inference, and faster training convergence. Experiments on benchmark PCB defect datasets demonstrate that our approach outperforms state-of-the-art methods in both accuracy and speed. The proposed model’s efficiency and high precision make it well suited for deployment in fast-paced PCB manufacturing lines that demand stringent real-time performance and reliable defect detection.

Kata kunci

Jaringan kolaborasi para penulis

8 penulis · 17 tautan kepenulisan bersama

Arahkan kursor atau ketuk sebuah nama untuk menyorot rekan penulisnya. Klik dua kali untuk membuka halaman penulis.

  • Industrial University of Ho Chi Minh City
  • Afiliasi tidak diketahui
  • Organisasi berbeda
  • Organisasi sama
  • Afiliasi tidak diketahui
  • Garis lebih tebal = lebih banyak artikel bersama · lingkaran lebih besar = lebih banyak artikel
Tampilkan sebagai tabel
Rekan penulisOrganisasiArtikel bersama
Tuan Anh NguyenAfiliasi tidak diketahui3
Thanh Quyen NgoIndustrial University of Ho Chi Minh City2
Long Ho LeAfiliasi tidak diketahui2
Trung Dung NguyenIndustrial University of Ho Chi Minh City2
Trung Kien PhamAfiliasi tidak diketahui2
Phan Minh ThanIndustrial University of Ho Chi Minh City1

Afiliasi berasal dari OpenAlex; metadata OAI-PMH jurnal tidak memuatnya. Penulis tanpa ID OpenAlex dicocokkan berdasarkan nama, sehingga dua orang bisa tergabung atau satu orang bisa terpisah.

Sumber metadata: OpenAlex API (jurnal SINTA, ADR 0048) https://openalex.org/S2765016208, pertama di-harvest 19 Nov 2026, terakhir terlihat 19 Nov 2026.
Metadata asli (oai_dc)
<oai_dc:dc xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:oai_dc="http://www.openarchives.org/OAI/2.0/oai_dc/"><dc:title xml:lang="en">Decoder-only transformer with multi-scale attention for efficient printed circuit board defect inspection</dc:title><dc:creator>Chi Kien Ha</dc:creator><dc:creator>Hoanh Nguyen</dc:creator><dc:subject>printed circuit board defect detection</dc:subject><dc:subject>decoder-only transformer</dc:subject><dc:subject>multi-scale attention</dc:subject><dc:subject>deformable attention</dc:subject><dc:subject>feature pyramid network</dc:subject><dc:subject>region proposal network</dc:subject><dc:subject>computational efficiency</dc:subject><dc:description xml:lang="en">Printed circuit boards (PCBs) are critical components in modern electronics, detecting defects rapidly, and accurately is crucial in manufacturing. We propose a novel real-time PCB defect detection model based on a Transformer decoder-only architecture. The framework first extracts multi-scale features via a backbone with a feature pyramid network (FPN) and generates candidate regions using a region proposal network (RPN). These region proposals are encoded as query embeddings that drive an adaptive multi-scale deformable attention (AMDA) module in the Transformer decoder, replacing the standard encoder-decoder attention. By dynamically weighting multi-scale feature maps for each query, AMDA emphasizes the feature scale most relevant to each defect’s size and texture, and yielding enhanced discriminative representations for subtle defect detection. The decoder-only design drastically reduces computational overhead compared to full encoder-decoder Transformers, enabling real-time inference, and faster training convergence. Experiments on benchmark PCB defect datasets demonstrate that our approach outperforms state-of-the-art methods in both accuracy and speed. The proposed model’s efficiency and high precision make it well suited for deployment in fast-paced PCB manufacturing lines that demand stringent real-time performance and reliable defect detection.</dc:description><dc:publisher>Institute of Advanced Engineering and Science (IAES)</dc:publisher><dc:date>2026-08-01</dc:date><dc:type>info:eu-repo/semantics/article</dc:type><dc:format>application/pdf</dc:format><dc:identifier>https://doi.org/10.11591/eei.v15i4.10873</dc:identifier><dc:identifier>10.11591/eei.v15i4.10873</dc:identifier><dc:source>Bulletin of Electrical Engineering and Informatics; Vol 15 No 4 (2026); 3070-3079</dc:source><dc:source>2089-3191</dc:source><dc:source>2302-9285</dc:source><dc:language>en</dc:language><dc:relation>https://beei.org/index.php/EEI/article/download/10873/4864</dc:relation><dc:rights>https://creativecommons.org/licenses/by-sa/4.0/</dc:rights></oai_dc:dc>