Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Rajdeep Chatterjee has a total of 72 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    IndoorCART: a spatially-aware transformer framework for Wi-Fi fingerprinting indoor localization

    by Balaram Singh, Rajdeep Chatterjee, Mahendra Kumar Gourisaria, Manoj Sahni, Ernesto León-Castro

    Accurate indoor localization remains difficult in GPS-denied environments because Wi-Fi received signal strength indicator (RSSI) fingerprints are sparse, noisy, device-dependent, and strongly affected by multipath propagation, non-line-of-sight conditions, and environmental dynamics. To address these challenges, we present IndoorCART (Indoor Context-Aware RSSI Transformer), an encoder-only transformer framework designed specifically for coordinate-regression-based Wi-Fi fingerprinting. IndoorCART represents access-point measurements as RSSI tokens, incorporates learnable access-point identity information, applies missing-signal-aware Adaptive Attention to model contextual dependencies among observed and unobserved signals, and combines CLS-token and global-average representations for robust coordinate prediction. The model is evaluated on three public indoor-localization benchmarks, namely, UJIIndoorLoc, TUJI1, and SODIndoorLoc, using a controlled comparison with nine reproducible regression baselines. IndoorCART achieves mean absolute errors of 4.79 m, 0.18 m, and 1.33 m on UJIIndoorLoc, TUJI1, and SODIndoorLoc, respectively, with corresponding R 2 scores of 0.99, 0.98, and 0.99. These results show that the proposed context-aware transformer consistently provides the lowest localization error among the implemented baselines and is particularly beneficial in sparse or structurally complex fingerprinting scenarios. We further position the proposed method against recent transformer-based indoor-localization studies through a literature-level comparison, while explicitly distinguishing reported external results from controlled reimplementations. Overall, IndoorCART offers an accurate and reliable framework for Wi-Fi-based indoor positioning; however, practical deployment should further consider model size, inference latency, energy consumption, device heterogeneity, and the modest performance gain observed on densely sampled datasets such as UJIIndoorLoc.

  • Open Access

    Towards Explainable and Robust Cervical Cancer Screening Using Domain-Specific Transfer Learning Algorithm

    by Jheelam Mondal, Mahendra Kumar Gourisaria, Rajdeep Chatterjee, Amitkumar V. Jha, Bhargav Appasani, Nicu Bizon, Cristian Toma

    Cervical cancer is the fourth most frequent malignancy in women globally. Pap smear screening is important for early cancer detection, but manual smear analysis is time-consuming, labor-intensive and error-prone for diagnosis. Such issues in resource-limited areas have led to the introduction of deep learning (DL) methods for automated cervical cancer diagnosis. But the majority of current methodologies depend on models pretrained on natural image datasets like ImageNet, which may inadequately represent domain-specific pathological characteristics. To mitigate this constraint, this research employs a domain-specific transfer learning algorithm approach using the PathMNIST histopathological dataset to enhance cervical cell classification. An accuracy score of 96.77% is achieved for the proposed YOLO* model, using the SIPAKMED dataset. To the best of available knowledge, no previous study has reported the use of PathMNIST as a pretraining source for cytology image classification. As domain-specific medical pretraining is becoming more popular, our study shows the importance of cross-domain generalization.

  • Open Access

    ERYXSeg: a hybrid CNN architecture for robust and resource-aware wound segmentation

    by Trishaani Acharjee, Rajdeep Chatterjee, Mahendra Kumar Gourisaria, Manoj Sahni, Ernesto León-Castro

    In modern clinical practice, the accuracy of wound segmentation is important because it allows for data-driven treatment planning, accurate evaluation of the wound area, and tracking of the healing process. In this paper, we introduce ERYXSeg, an integrated deep learning architecture that combines residual skip connections, a parameter-efficient core, and spatial localization and boundary-aware refinement of a segmentation head. The proposed method achieves robust performance across a wide range of wound types by capturing both local fine-grained characteristics and rich contextual information. ERYXSeg outperforms the current state-of-the-art models under identical training settings after being trained from scratch on mixed wound datasets. It achieves the highest reported IoU and Dice scores in both datasets: 0.7633 IoU and 0.8658 Dice for the foot ulcer dataset and 0.6910 IoU and 0.8173 Dice for the curated mixed wound dataset. The contribution of each architectural element is further validated by a systematic ablation study, which shows that attention-gated skip connections are crucial for accurate spatial reconstruction across various data sources and that Mobile Inverted Bottleneck Convolution (MBConv) blocks are essential for extracting broadly applicable features. The results obtained demonstrate excellent generalization and high computational efficiency of the model, which thus qualifies it for real-time clinical deployment independent of transfer learning.


  • ERYXNet: a lightweight yet robust architecture for multi-type wound classification

    by Trishaani Acharjee, Rajdeep Chatterjee, Biswaranjan Acharya, Fernando Moreira

  • Drone-Based Parking Occupancy Monitoring and Recommendation

    by Rajdeep Chatterjee, Subhranil Roy, Sourov Roy Shuvo, Rajesh Chowdhury, Sorup Chakraborty

  • Automatic Classification of Cervical Dysplasia Using Deep Neural Network

    by Jheelam Mondal, Rajdeep Chatterjee, Mahendra Kumar Gourisaria

  • Open Access

    LPM: 3D Lungs Reconstruction using Lung Parametric Model

    by Shreya Chakraborty, Tathagata Bandyopadhyay, Rajdeep Chatterjee

    Deep learning techniques have been proven fruitful for many researchers in the computer vision field for problems like image segmentation and object recognition. With the success of this, 3D reconstruction has been of great interest, and how Deep Learning can be used to implement the same. There have been plenty of successfully developed models and methods that take Computer Tomography (CT) scans as an input, then preprocessed to obtain the specific organ (lungs, liver, heart, etc.) and offer a stacked 3D representation of the organ as output. This paper reviews some of the recent papers that offer a different approach on how a 3D object can be constructed from an Xray image alone, along with the challenges that come with such. This paper will provide insights into the development of a model capable of reconstructing 3D lung structures from point cloud data, utilizing real-world datasets. Additionally, it will explore the application of latent code for generating a broader dataset of lung models. This approach has the potential to aid in identifying the size and shape of lungs directly from X-ray images.

  • Beyond Microscope: Advances in Technology for Detection of Cervical Cancer

    by Prasun Payne, Rohi Velgina Romould, Jheelam Mondal, Rajdeep Chatterjee, Mahendra Kumar Gourisaria


  • Open Access

    Temporal-Aware Transformer Approach for Violence Activity Recognition

    by Rajdeep Chatterjee, Ritabrata Roy Choudhury, Mahendra Kumar Gourisaria, Sreejata Banerjee, Soumik Dey, Manoj Sahni, Ernesto León-Castro

    The need for effective violence detection in public spaces has intensified with increasing antisocial behavior and violence. Traditional surveillance systems, which are relying on human operators, face delays and resource challenges. Using advances in artificial intelligence (AI) and computer vision, this research presents a scalable deep learning architecture for real-time violence detection using two approaches. In the first approach, Convolutional Neural Networks (CNN) and bidirectional long-short-term memory (BiLSTM) networks are combined, where MobileNetV2 is used for spatial feature extraction and BiLSTM for temporal pattern recognition, achieving an accuracy of 95.6%. The second approach incorporates a spatial-temporal transformer (TransformerSeq) in place of BiLSTM, improving performance to 97.2% by capturing spatiotemporal relationships in video data more effectively through self-attention for temporal feature learning. The lightweight SOTA MobileNetV2, along with the proposed MobileTransformerSeq, enables the effective differentiation between violent and non-violent activities, demonstrating the potential to enhance public safety in diverse settings.