IAIrfan Abbas

Irfan Abbas

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Irfan Abbas has a total of 11 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    UrbanTreeFormer: A boundary-aware CNN-transformer architecture for fine-grained urban street tree segmentation

    by Irfan Abbas, Rytis Maskeliūnas, Sarmad Maqsood, Ali Amini, Izabela Karsznia, Robertas Damasevicius

    Accurate segmentation of urban street trees from street-level imagery supports tree inventory, ecological assessment, biodiversity monitoring, and smart-city planning. However, irregular crown boundaries, fragmented foliage, occlusion, background clutter, and class imbalance remain significant challenges. This paper presents UrbanTreeFormer, a boundary-aware CNN–Transformer architecture for fine-grained urban tree crown segmentation. The model integrates an EfficientNet-B3 encoder for local feature extraction, a Transformer bottleneck for long-range contextual modeling, an attention-gated decoder for selective feature refinement, and an auxiliary boundary-refinement head to improve crown delineation. A combined weighted cross-entropy, Dice, and boundary loss, alongside class weighting and oversampling, addresses the foreground–background imbalance and the moderate imbalance among species. UrbanTreeFormer was evaluated on the public Urban Street Tree Segmentation dataset comprising 3,949 images and 22 tree species, using binary foreground and species-wise multiclass segmentation protocols. On the 386-image held-out test set, binary segmentation achieved Accuracy of 0.9731 ± 0.0014, Precision of 0.8695 ± 0.0032, Recall of 0.9544 ± 0.0039, Dice of 0.9100 ± 0.0021, and IoU of 0.8349 ± 0.0033. Boundary evaluation yielded BF1 of 0.8810 ± 0.0027, Boundary IoU of 0.7890 ± 0.0040, ASSD of 2.08 ± 0.08 pixels, and HD95 of 6.95 ± 0.21 pixels. Species-wise evaluation achieved macro-averaged Dice and IoU scores of 0.8731 and 0.7778, respectively. UrbanTreeFormer achieved the highest mean performance among eight CNN-, attention-, and Transformer-based baselines, with statistically significant improvements over seven, although its difference from SegFormer was not statistically significant. Ablation studies verified the contribution of each architectural component, while robustness experiments demonstrated stability under moderate photometric variations but reduced performance under severe noise, blur, and compression. These findings highlight UrbanTreeFormer's potential for accurate urban tree crown segmentation and species-level monitoring.

  • Open Access

    Multi-Scale Spatiotemporal EEG and Self-Supervised Audio Fusion: A Mixture-of-Experts Approach to Continuous Affect

    by Ali Amini, Sarmad Maqsood, Irfan Abbas, Muhammad Abdullah Sarwar, Rytis Maskeliūnas, Robertas Damaševičius

  • Open Access

    Simulation-Based Evaluation of LiDAR–Photogrammetry Fusion via NeRF Reconstruction and ICP Registration in Urban Scenes

    by Rytis Maskeliūnas, Sarmad Maqsood, Ahmad Qurthobi, Irfan Abbas, Mantas Vaškevičius, Julius Gelšvartas

    Reliable integration of LiDAR and photogrammetric point clouds is essential for high-precision 3D mapping, yet systematic evaluations of fusion accuracy under controlled conditions remain limited. This study presents a simulation-based assessment framework for LiDAR–photogrammetry fusion using a neural radiance field (NeRF) representation and iterative closest point (ICP) registration. Experiments were conducted in the CARLA simulator (Town 10), where a drone-mounted multi-sensor platform (RGB, LiDAR, GPS, IMU) generated spatially aligned datasets. Photogrammetric reconstructions were produced using the Nerfstudio–Nerfacto pipeline with varied architectural and sampling configurations to analyze their impact on cross-modal registration. Quantitative evaluation employed Chamfer and cloud-to-cloud (C2C) distances to assess NeRF reconstruction fidelity, and ICP root-mean-square error (RMSE), inlier fitness, and runtime to evaluate registration performance. Results show that tuning NeRF’s hidden dimensions and sampling levels yields up to 25% lower ICP RMSE and faster convergence across object categories. The proposed framework enables reproducible benchmarking of LiDAR–photogrammetry fusion and provides a foundation for extending NeRF-based methods to real-world urban mapping scenarios.