Internship - Low quality barcode decoding
Description courte
At IDEMIA Public Security, we support governments and public institutions with advanced identity and security solutions.
Our teams develop technologies that help build a safer world by protecting people, assets, and identities.
IDEMIA develops high-performance algorithms to authenticate Identity documents. One of them is the barcoding decoding algorithm. Barcodes are useful to check the consistency between the text written on the document and the one encoded in the barcode. In case of low-quality barcode, the percentage of failures is more important. Low-quality means blurred image, low resolution image or badly printed barcodes. The purpose of the internship is to design, implement and evaluate deep learning algorithms to improve the percentage of decoded barcodes. It can be restoration algorithms, binarization algorithms, super-resolution algorithms, … or the combination of these algorithms.
Key Missions
- Design, implement and evaluate deep learning algorithms to improve the percentage of decoded barcodes
- Develop deep learning models to enhance degraded barcode images
- Build and preprocess datasets with real and synthetic degradations
- Train and evaluate models in PyTorch using advanced vision architectures
Profile & Other Information
Education
- Engineering student from a prestigious school in the final year of the engineering cycle, or a student in a Master's program, specializing in deep learning, computer vision and image processing.
- Required technical skills
- Strong background in image processing, computer vision, and deep learning (ideally with PyTorch).
Soft skills & languages
- Curious. Proactive, autonomous.
- Fluent spoken and written English
The internship will last 6 months, starting in 2026, and will be based in Osny, Val d’Oise, France.
The intern will integrate a team of AI scientists who work on algorithms to authenticate Identity Documents.