Image-Based Intelligent Classroom Monitoring and Learning Behavior Analysis via Joint Object Detection and Semantic Segmentation
Bibliographic record
Abstract
With the advancement of educational informatization, intelligent classrooms increasingly rely on image analysis technologies to automate environmental monitoring and learning behavior analysis.However, current research faces three key limitations: (1) the disjoint handling of object detection and semantic segmentation leads to suboptimal feature utilization; (2) existing models perform poorly in detecting dim classroom boundaries due to inadequate attention mechanisms; and (3) conventional loss functions struggle to address the pixel imbalance between boundary lines and the background.To address these challenges, this paper proposes a joint object detection and semantic segmentation model tailored for intelligent classroom scenarios.The model employs a shared encoder with dual decoder branches to achieve collaborative reasoning for both environmental object detection and learning behavior region segmentation.A Bi-directional Feature Pyramid Network (BiFPN) is integrated to introduce an attention-like weighted feature fusion mechanism, enhancing the capture of subtle boundary features.Additionally, an improved EFL Focal Loss is introduced to mitigate pixel imbalance issues.The main contributions of this work include: constructing a unified framework to enhance feature synergy between detection and segmentation tasks; designing a targeted attention mechanism to optimize boundary detection; and improving the loss function to balance pixel-wise training.Experimental results demonstrate improved completeness and accuracy in classroom scene analysis.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".