Boutaina Hdioud

Prof. Boutaina Hdioud


Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 June 2023 in IAES International Journal of Artificial Intelligence (IJ-AI)
Verified
Journal Article
Open access
PDF file
BHBoutaina Hdioud
Boutaina Hdioud
MTMohammed El Haj Tirari
Mohammed El Haj Tirari
Facial expression recognition (FER) represents one of the most prevalent forms of interpersonal communication, which contains rich emotional information. But it became even more challenging during the times of COVID, where face masks became a mandatory protection measure, leading to the challenge of occluded lower-face during facial expression recognition. In this study, deep convolutional neural network (DCNN) represents the core of both our full-face FER system and our masked face FER model. The focus was on incorporating knowledge distillation in transfer learning between a teacher model, which is the full-face FER DCNN, and the student model, which is the masked face FER DCNN via the combination of both the loss from the teacher soft-labels vs the student soft labels and the loss from the dataset hard-labels vs the student hard-labels. The teacher-student architecture used FER2013 and a masked customized version of FER2013 as datasets to generate an accuracy of 69% and 61% respectively. Therefore, the study proves that the process of knowledge distillation may be used as a way for transfer learning and enhancing accuracy as a regular DCNN model (student only) would result in 46% accuracy compared to our approach (61% accuracy).

8 October 2025 in Information
Verified
Journal Article
Open access
PDF file
SHSohaya EL HAMDOUNI
Sohaya EL HAMDOUNI
BHBoutaina Hdioud
Boutaina Hdioud
SFSanaa EL FKIHI
Sanaa EL FKIHI
Object detection has a vital impact on the analysis and interpretation of visual scenes. It is widely utilized in various fields, including healthcare, autonomous driving, and vehicle surveillance. However, complex scenes containing small, occluded, and multiscale objects present significant difficulties for object detection. This paper introduces a lightweight object detection algorithm, utilizing YOLOv8n as the baseline model, to address these problems. Our method focuses on four steps. Firstly, we add a layer for small object detection to enhance the feature expression capability of small objects. Secondly, to handle complex forms and appearances, we employ the C2f-DCNv2 module. This module integrates advanced DCNv2 (Deformable Convolutional Networks v2) by substituting the final C2f module in the backbone. Thirdly, we designed the CBAM, a lightweight attention module. We integrate it into the neck section to address missed detections. Finally, we use Ghost Convolution (GhostConv) as a light convolutional layer. This alternates with ordinary convolution in the neck. It ensures good detection performance while decreasing the number of parameters. Experimental performance on the PASCAL VOC dataset demonstrates that our approach lowers the number of model parameters by approximately 9.37%. The mAP@0.5:0.95 increased by 0.9%, recall (R) increased by 0.8%, mAP@0.5 increased by 0.3%, and precision (P) increased by 0.1% compared to the baseline model. To better evaluate the model’s generalization performance in real-world driving scenarios, we conducted additional experiments using the KITTI dataset. Compared to the baseline model, our approach yielded a 0.8% improvement in mAP@0.5 and 1.3% in mAP@0.5:0.95. This result indicates strong performance in more dynamic and challenging conditions.

26 September 2025 in Information
Verified
Journal Article
Open access
PDF file
MEMERIEME ELKORCHI
MERIEME ELKORCHI
BHBoutaina Hdioud
Boutaina Hdioud
OROULAD HAJ THAMI Rachid
OULAD HAJ THAMI Rachid
SMSafae Merzouk
Safae Merzouk
In recent years, most conventional emotion recognition approaches have concentrated primarily on facial cues, often overlooking complementary sources of information such as body posture and contextual background. This limitation reduces their effectiveness in complex, real-world environments. In this work, we present a multi-branch emotion recognition framework that separately processes facial, bodily, and contextual information using three dedicated neural networks. To better capture contextual cues, we intentionally mask the face and body of the main subject within the scene, prompting the model to explore alternative visual elements that may convey emotional states. To further enhance the quality of the extracted features, we integrate both channel and spatial attention mechanisms into the network architecture. Evaluated on the challenging NCAER-S dataset, our model achieves an accuracy of 56.42%, surpassing the state-of-the-art GLAMOUR-Net. These results highlight the effectiveness of combining multi-cue representation and attention-guided feature extraction for robust emotion recognition in unconstrained settings. The findings also highlight the importance of accurate emotion recognition for human–computer interaction, where affect detection enables systems to adapt to users and deliver more effective experiences.

26 August 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
YZYassine Zouhdi
Yassine Zouhdi
BHBoutaina Hdioud
Boutaina Hdioud
The advancements in the reasoning abilities of large language models (LLMs) have been almost exclusively based upon chain-of-thought, which improve performance but increase the amount of verbose tokens within each response. As a result, the increased output size produces both slower inference and increased memory usage. These two factors provide a substantial barrier to deployment in limited resource environments as those commonly found in edge computing and robotics. This paper provides an implementation of ”cognitive density” to help bridge the gap between reasoning complexity and computational resources. We applied group relative policy optimization (GRPO), a Reinforcement Learning method that eliminates the need for a memory-intensive value network, to align a 4 billion parameter small language model (SLM) to produce mathematically correct responses. The goal is to resolve complex problems while being constrained by generating less than 250 total tokens. By designing a constraint aware reward structure we incentivized the model to generate highly concise reasoning. Our research identifies significant shortcomings of the present architectures; when using a token limit of 250 tokens, a state of the art Instruct SLM is reduced to 76% accuracy whereas its Thinking version is catastrophically decreased to 10% accuracy. This is due to current reasoning models having structural dependence on output length to produce coherent logic. In contrast Our aligned model performed at nearly 87% accuracy while being similarly constrained. Additionally, these results also showed that training a model for dense thinking was computationally efficient and reached convergence in under 1 hour on one GPU.


20 June 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
FLFatiha Limami
Fatiha Limami
BHBoutaina Hdioud
Boutaina Hdioud
OROULAD HAJ THAMI Rachid
OULAD HAJ THAMI Rachid
ANAssia Naja
Assia Naja
Facial expression recognition remains a core challenge in computer vision, but it's becoming more important in security, adaptive learning, mental health evaluation, and interactive systems. Deep learning models work well, but genuine progress is still restricted by the fact that existing datasets typically don't show the full range and complexity of real-world emotional expressions. A lot of them had unbalanced emotion classifications, not enough variety in context, and were collected in very controlled settings that don't reflect how things are in the real world. A lot of benchmarks just look at Ekman's six main feelings. This limits the range and difficulty that models need to work well in real life. To solve these difficulties, we provide EMOTIF, a collection of real-world facial expressions that shows 15 different emotions in diverse settings. Using advanced detection and annotation methods, an automated processing system created accurate and full emotion data. The FYOLOv5 model (Face: You Only Look Once) is a fast, first-pass detector that performs well for identifying faces in real time and in tough settings. This is basically what it does. Because of this, YOLOv11, a more complex model, makes bounding box predictions better, making sure that face areas are retrieved correctly and with a lot of information. This hybrid detection method creates a well-organized, full dataset that is a great starting point for building reliable, context-aware face emotion recognition algorithms and for making affective computing applications work better in real life.