CENTRALE LYON - Post-doctorant en vision et apprentissage : Détection automatique de défauts thermiques dans des prises de vues intérieures de façades pour l’aide au diagnostic de qualité des bâtiments

ECL and Laboratory presentation

Founded in 1857, École Centrale de Lyon is one of the top 10 engineering schools in France. It trains more than 3,000 students of 50 different nationalities on its campuses in Écully and SaintÉtienne (ENISE, in-house school): general engineers, specialized engineers, masters and

doctoral students. With the Groupe des Écoles Centrale, it has three international locations.

The training provided benefits from the excellence of the research carried out in the 6 CNRS accredited laboratories on its campuses, the 2 international laboratories, the 6 international research networks and the 10 joint laboratories with companies. Its excellent research and

high-level teaching have enabled it to establish double degree agreements with prestigious universities and advanced partnerships with numerous companies. With its focus on sobriety, energy, the environment and decarbonization, Centrale Lyon intends to respond to the

problems faced by socio-economic players in the major transitions.

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

Research field presentation :

This mission is part of the Therm_I_Bat project, a collaboration between LIRIS and the company CMDL Manaslut Ing., which aims to develop new AI and Computer Vision methods for the automatic detection of thermal defects in interior façade imagery to assist

in building quality diagnostics.

Mission Description

The actual energy performance of new or renovated buildings often falls short of initial design targets, creating an "energy gap" that slows down the sector's decarbonization. CMDL has developed a robotic solution capable of conducting comprehensive measurement campaigns (infrared thermography and RGB imagery) across office floors, housing units, or industrial plants covering roughly 5,000 m² in a single night. However,

data processing is currently handled manually by an expert on a limited number of files.

With the growing volume of files, relying on expert review is no longer feasible, making automated processing based on Artificial Intelligence and machine learning the only viable path forward.

The THERM_I_BAT project aims to design and validate a proof of concept for intelligent building façade diagnostics, using interior imagery captured by a robotic platform. Scientifically, the project aligns with research and R&D efforts in multimodal image analysis

and frugal machine learning by developing new AI models capable of operating on limited yet highly contextualized datasets. Industrially, it prepares an automated diagnostic solution for building envelope quality that is compatible with existing industry tools (BIM,

digital twins) and cost-effective for large-scale deployment upon building handover.

The projected 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) across several pilot sites.

• Assembly of an annotated dataset (defects, sound areas, BIM context).

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

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

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

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

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

• Integration of CMDL expert knowledge to guide classification 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 using limited datasets, potentially supplemented by simulated data.

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

• Automated generation of structured diagnostics (defect inventory, typology, severity level) and spatial mapping within 2D plans/BIM.

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

• Consolidation of the proof of concept (TRL 5) and preparation for subsequent commercialization and deployment steps.

Required skills / qualifications

Diplomas : PhD diploma for the post-doc position, Master 2 for the Research Engineer position

Knowledge required: Computer Vision, Machine Learning, Deep Learning

Operational skills : Good proficiency in the Python language and deep learning libraries such as PyTorch, proficiency in English, and good written and oral communication skills.

Behavioural skills : Motivation, rigor, autonomy, proactivity..

Recruitment process

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

• Study of the written application: CV + cover letter + PhD diploma

• Selection interview: in person or by videoconference

Deadline for application : November 8, 2026