CENTRALE LYON - Postdoctoral Researcher in Vision and Learning: Automatic Detection of Thermal Defects in Interior Façade Views for Building Quality Diagnosis Assistance

Presentation of the School and the Laboratory

Founded in 1857, École Centrale de Lyon is among the top 10 engineering schools in France. It trains over 3,000 students from 50 different nationalities on its Écully and Saint-Étienne campuses (ENISE, an internal school): generalist engineers, specialized engineers, masters, and doctoral students. With the Groupe des Écoles Centrale, it has 3 international locations. The education provided benefits from the excellence of research from the 6 CNRS-labeled laboratories on its campuses, 2 international laboratories, 6 international research networks, and 10 joint laboratories with companies. Its excellent research and very high-level teaching enable it to establish double degree agreements with prestigious universities and cutting-edge partnerships with numerous companies. Focusing on themes of sobriety, energy, environment, and decarbonization, Centrale Lyon aims to address the challenges of socio-economic actors in major transitions.

This research project will be conducted within the LIRIS Laboratory (UMR 5205), and specifically within the Imagine team, whose work focuses on developing new learning and vision models for the analysis and understanding of visual data such as images and videos.

Presentation of the Topic:

This mission is part of the Therm_I_Bat project, a collaboration between LIRIS and the company CMDL Manaslut Ing., with the objective of proposing new AI and Computer Vision methods for the automatic detection of thermal defects in interior façade views to assist in building quality diagnosis.

Description of the Mission

The actual energy performance of new or renovated buildings often falls short of the performance targeted during design, creating an "energy gap" that hinders the decarbonization of the sector. CMDL has developed a robotic solution capable of conducting exhaustive measurement campaigns (infrared thermography and RGB images) on office floors, housing, or factory spaces of approximately 5,000 m² in one night. However, currently, the processing is performed by an expert on a limited number of files. The multiplication of files makes expert mobilization unfeasible, and only automated processing based on Artificial Intelligence-type learning is conceivable.

The THERM_I_BAT project aims to design and validate a proof of concept for intelligent building façade diagnosis, based on interior views captured by a robotic platform.

Scientifically, the project is part of research and R&D efforts on multimodal image analysis and frugal learning, by developing new AI models capable of operating with limited but highly contextualized datasets. Industrially, it prepares an automated quality diagnosis solution for building envelopes, compatible with industry tools (BIM, digital twin) and economically viable for massive deployment upon building reception.

The planned schedule is as follows:

Phase 1 – Data Acquisition and Preparation (M1–M12)

• Definition of measurement protocols with CMDL (building typologies, defect scenarios).

• Execution of robotic measurement campaigns (IR + RGB) on several pilot sites.

• Creation of an annotated dataset (defects, healthy areas, BIM context).

Phase 2 – Spatial Contextualization and Multimodal Fusion (M3–M5)

• Spatial alignment of IR and RGB images, considering façade geometry.

• Development of multimodal fusion methods to produce robust joint representations.

Phase 3 – Semantic Segmentation and Perturbation Filtering (M6–M9)

• Development of semantic segmentation models to distinguish relevant areas (walls, joinery, technical elements) from perturbations (furniture, objects, reflections).

• Integration of CMDL expert knowledge to guide classes and filtering rules.

Phase 4 – Anomaly Detection and Defect Classification (M10–M16)

• Implementation of thermal anomaly detection and defect classification models (thermal bridges, air leakage defects, material-related artifacts).

• Frugal learning from limited datasets, possibly supplemented by simulated data.

Phase 5 – Diagnosis Generation and Validation in a Relevant Environment (M16–M18)

• Automatic generation of structured diagnoses (list of defects, typology, severity) and localization in the plan/BIM.

• Validation campaigns on new buildings, comparison with human expert analysis (time, accuracy, false positive/negative rates).

• Consolidation of the proof of concept (TRL 5) and preparation for further development.

Profile Sought / Expected Skills

Degree: Doctorate in Computer Science

Required Knowledge: Computer Vision, Machine Learning, Deep Learning

Operational Skills: Good command of Python and deep learning libraries such as PyTorch, English proficiency, and good written and oral communication skills

Behavioral Skills: Motivation, rigor, autonomy, proactivity.

The recruitment process consists of two stages, supervised by a recruitment committee, in accordance with Centrale Lyon's OTMR policy.

· Study of the written application: CV + cover letter + doctorate diploma

· Selection interview: in-person or via videoconference

Application Deadline: November 8, 2026

Find Jobs in France on Arbeitnow