{"id":13911,"date":"2025-11-16T12:30:21","date_gmt":"2025-11-16T11:30:21","guid":{"rendered":"https:\/\/www.ibisc.univ-evry.fr\/salah-eddine-laidoudi-will-defend-his-doctoral-thesis-on-monday-november-17-2025-deep-learning-for-immersion-and-natural-3d-interaction-within-intelligent-mixed-reality-experiences\/"},"modified":"2025-11-19T18:14:24","modified_gmt":"2025-11-19T17:14:24","slug":"salah-eddine-laidoudi-will-defend-his-doctoral-thesis-on-monday-november-17-2025-deep-learning-for-immersion-and-natural-3d-interaction-within-intelligent-mixed-reality-experiences","status":"publish","type":"post","link":"https:\/\/www.ibisc.univ-evry.fr\/en\/salah-eddine-laidoudi-will-defend-his-doctoral-thesis-on-monday-november-17-2025-deep-learning-for-immersion-and-natural-3d-interaction-within-intelligent-mixed-reality-experiences\/","title":{"rendered":"Salah-eddine LAIDOUDI will defend his doctoral thesis on Monday, November 17, 2025: “Deep Learning for Immersion and Natural 3D Interaction within Intelligent Mixed Reality Experiences”"},"content":{"rendered":"

<\/div><\/div>
\n\t\t\t\t\t
\n\t\t\t\t\t\t

Salah-eddine LAIDOUDI defends his doctoral thesis on thirsday, November 17, 2025, 2pm, University of \u00c9vry, Pelvoux Site, Yasmina Bestaoui Bx30 Amphitheather. The Thesis defense is aso accessible via Zoom: https:\/\/univ-evry-fr.zoom.us\/j\/92115017016?pwd=Kes1az5N9Bpz9WjOZPOn8daYFbpdWJ.1<\/a><\/strong><\/span> .<\/p>\n

Title : Deep Learning for Immersion and Natural 3D Interaction within Intelligent Mixed Reality Experiences.<\/h2>\n

Keywords<\/h2>\n

mixed reality, 3D interaction, deep learning, object detection, AR<\/p>\n

Abstract<\/h2>\n
\n

Mixed Reality aims to merge real and virtual worlds so that digital objects feel present and directly manipulable. Two hurdles persist: real-time scene understanding at 30\u201360 fps and natural 3D interaction with bare hands under tight latency and power constraints. Recent anchor-free transformers, multi-scale attention, and temporal-shift networks are accurate but often exceed the 30 ms budget on standalone devices. This thesis asks how to co-design models, data, and deployment so a mobile MR device achieves reliable detection and hands-free interaction within strict real-time (< 30 ms) and compute (\u2248 10 GFLOPs) envelopes, since compression alone rarely suffices.<\/p>\n

We introduce SSAFT (Single-Shot Anchor-Free Transformer), coupling a shallow CNN backbone with a compact transformer decoder (7.8 M parameters; 7.3 GFLOPs) trained on Indoor-Objects-28 (19,162 images, 28 classes). To better capture small and low-texture targets under the same budget, we add DFMA (Dual-Focus Multi-Scale Attention), a depthwise 3\/5\/7-pixel block with squeeze\u2013excite gating inserted at three stages for \u2264 5% FLOPs overhead. To probe robustness, a synthetic astronomy corpus (40,000 renders) enables sim-to-real adaptation, yielding +3 AP on real astrophotographic images without violating runtime constraints.<\/p>\n

For hand interaction, dense 3D networks (I3D, SlowFast) reach high accuracy but need tens of GFLOPs and large memory; lightweight temporal-shift variants improve efficiency yet plateau near 90% Top-1 accuracy. We therefore design G-DiTSM, a compact MobileNet-based temporal-shift model that isolates motion via forward and backward frame differences, trained end-to-end on 20BN-Jester (148k clips, 27 gestures). It preserves around 95% Top-1 accuracy, fits in approximately 15 MB, and runs comfortably above 30 fps on mobile devices with sub-10 ms latency.<\/p>\n

Under fair settings, Indoor-Objects-28 results show SSAFT at 42.0% mAP50:95 with 12 ms average inference; YOLOv8-n attains 40.4% at 13 ms, while RT-DETR reaches 45% but is about 15\u00d7 slower (184 ms). On MS-COCO, adding DFMA lifts mAP50:95 to 41.3% at 7.6 GFLOPs\u2014about +4 points at a runtime comparable to YOLOv8-n\u2014 whereas RT-DETR-R18 (46.5%) exceeds the target compute (~15 GFLOPs). For gestures, G-DiTSM achieves 95.3% Top-1 accuracy with just 0.084 GFLOPs and < 10 ms latency, outperforming MobileNet-TSM by 5.5% while using an order of magnitude fewer resources than 3D baselines.<\/p>\n

After INT8 quantization and Unity\/Sentis integration, the full stack sustains continuous real-time throughput on consumer XR hardware; a user study indicates corresponding gains in perceived speed and usability. Looking ahead, RGB\u2013depth\u2013IMU fusion, CLIP-style open-set detection, continuous-stream gesture segmentation, LLM-based context, and adaptive INT8\/FP16 switching offer principled paths to broader robustness while keeping compute and energy budgets in check.<\/p>\n<\/div>\n

<\/div>\n<\/div>
<\/div>
<\/div>

Composition du jury de th\u00e8se\/Doctoral thesis jury composition<\/h3>\n<\/div>\n
\n\n\n\n\n\n\n\n\n\n\n
Membre du jury<\/th>\nTitre<\/th>\nLieu d’exercice<\/th>\nFonction dans le jury<\/th>\n<\/tr>\n<\/thead>\n
Mehdi AMMI<\/td>\nProfesseur des Universit\u00e9s<\/td>\nUniversit\u00e9 Paris 8<\/td>\nExaminateur<\/td>\n<\/tr>\n
Hanane AZZAG<\/td>\nProfesseure des Universit\u00e9s<\/td>\nUniversit\u00e9 Sorbonne Paris Nord<\/td>\nExaminatrice<\/td>\n<\/tr>\n
Elhadj BENKHELIFA<\/td>\nFull Professor<\/td>\nStaffordshire University<\/td>\nRapporteur<\/td>\n<\/tr>\n
Madjid MAIDI<\/td>\nMa\u00eetre de Conf\u00e9rences<\/td>\nUniversit\u00e9 Paris 8<\/td>\nCo-encadrant de th\u00e8se<\/td>\n<\/tr>\n
Samir OTMANE<\/td>\nProfesseure des Universit\u00e9s<\/td>\nUniversit\u00e9 \u00c9vry Paris-Saclay<\/td>\nDirecteur de th\u00e8se<\/td>\n<\/tr>\n
Titus ZAHARIA<\/td>\nProfesseur<\/td>\nInstitut Polytechnique de Paris \/ TELECOM SudParis<\/td>\nRapporteur<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n
<\/div>
<\/div>