Pooja Guhan

Pooja Guhan

Auto-generated profile, based on publicly available information

Pooja Guhan

University of Maryland College Park

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

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

Representative Publications

View All Publications
14 March 2020 in arXiv
Verified
Preprint
Open access
PDF file
TMTrisha Mittal
Trisha Mittal
PGPooja Guhan
Pooja Guhan
UBUttaran Bhattacharya
Uttaran Bhattacharya
RCRohan Chandra
Rohan Chandra
ABAniket Bera
Aniket Bera
DMDinesh Manocha
Dinesh Manocha
We present EmotiCon, a learning-based algorithm for context-aware perceived human emotion recognition from videos and images. Motivated by Frege's Context Principle from psychology, our approach combines three interpretations of context for emotion recognition. Our first interpretation is based on using multiple modalities(e.g. faces and gaits) for emotion recognition. For the second interpretation, we gather semantic context from the input image and use a self-attention-based CNN to encode this information. Finally, we use depth maps to model the third interpretation related to socio-dynamic interactions and proximity among agents. We demonstrate the efficiency of our network through experiments on EMOTIC, a benchmark dataset. We report an Average Precision (AP) score of 35.48 across 26 classes, which is an improvement of 7-8 over prior methods. We also introduce a new dataset, GroupWalk, which is a collection of videos captured in multiple real-world settings of people walking. We report an AP of 65.83 across 4 categories on GroupWalk, which is also an improvement over prior methods.

26 October 2024
Verified
Book Chapter
Closed access
PDF file
PGPooja Guhan
Pooja Guhan
THTsung-Wei Huang
Tsung-Wei Huang
GSGuan-Ming Su
Guan-Ming Su
SGSubhadra Gopalakrishnan
Subhadra Gopalakrishnan
DMDinesh Manocha
Dinesh Manocha

20 January 2025 in JMIR Formative Research
Verified
Journal Article
Open access
PGPooja Guhan
Pooja Guhan
NANaman Awasthi
Naman Awasthi
KMKathryn McDonald
Kathryn McDonald
KBKristin Bussell
Kristin Bussell
GRGloria Reeves
Gloria Reeves
ABAniket Bera
Aniket Bera
Background: Patient engagement is a critical but challenging public health priority in behavioral health care. During telehealth sessions, health care providers need to rely predominantly on verbal strategies rather than typical nonverbal cues to effectively engage patients. Hence, the typical patient engagement behaviors are now different, and health care provider training on telehealth patient engagement is unavailable or quite limited. Therefore, we explore the application of machine learning for estimating patient engagement. This can assist psychotherapists in the development of a therapeutic relationship with the patient and enhance patient engagement in the treatment of mental health conditions during tele–mental health sessions. Objective: This study aimed to examine the ability of machine learning models to estimate patient engagement levels during a tele–mental health session and understand whether the machine learning approach could support therapeutic engagement between the client and psychotherapist. Methods: We proposed a multimodal learning-based approach. We uniquely leveraged latent vectors corresponding to affective and cognitive features frequently used in psychology literature to understand a person’s level of engagement. Given the labeled data constraints that exist in health care, we explored a semisupervised learning solution. To support the development of similar technologies for telehealth, we also plan to release a dataset called Multimodal Engagement Detection in Clinical Analysis (MEDICA). This dataset includes 1229 video clips, each lasting 3 seconds. In addition, we present experiments conducted on this dataset, along with real-world tests that demonstrate the effectiveness of our method. Results: Our algorithm reports a 40% improvement in root mean square error over state-of-the-art methods for engagement estimation. In our real-world tests on 438 video clips from psychotherapy sessions with 20 patients, in comparison to prior methods, positive correlations were observed between psychotherapists’ Working Alliance Inventory scores and our mean and median engagement level estimates. This indicates the potential of the proposed model to present patient engagement estimations that align well with the engagement measures used by psychotherapists. Conclusions: Patient engagement has been identified as being important to improve therapeutic alliance. However, limited research has been conducted to measure this in a telehealth setting, where the therapist lacks conventional cues to make a confident assessment. The algorithm developed is an attempt to model person-oriented engagement modeling theories within machine learning frameworks to estimate the level of engagement of the patient accurately and reliably in telehealth. The results are encouraging and emphasize the value of combining psychology and machine learning to understand patient engagement. Further testing in the real-world setting is necessary to fully assess its usefulness in helping therapists gauge patient engagement during online sessions. However, the proposed approach and the creation of the new dataset, MEDICA, open avenues for future research and the development of impactful tools for telehealth.

14 January 2025 in arXiv
Verified
Preprint
Open access
PDF file
PGPooja Guhan
Pooja Guhan
THTsung-Wei Huang
Tsung-Wei Huang
GSGuan-Ming Su
Guan-Ming Su
SGSubhadra Gopalakrishnan
Subhadra Gopalakrishnan
DMDinesh Manocha
Dinesh Manocha
We introduce V-Trans4Style, an innovative algorithm tailored for dynamic video content editing needs. It is designed to adapt videos to different production styles like documentaries, dramas, feature films, or a specific YouTube channel's video-making technique. Our algorithm recommends optimal visual transitions to help achieve this flexibility using a more bottom-up approach. We first employ a transformer-based encoder-decoder network to learn recommending temporally consistent and visually seamless sequences of visual transitions using only the input videos. We then introduce a style conditioning module that leverages this model to iteratively adjust the visual transitions obtained from the decoder through activation maximization. We demonstrate the efficacy of our method through experiments conducted on our newly introduced AutoTransition++ dataset. It is a 6k video version of AutoTransition Dataset that additionally categorizes its videos into different production style categories. Our encoder-decoder model outperforms the state-of-the-art transition recommendation method, achieving improvements of 10% to 80% in Recall@K and mean rank values over baseline. Our style conditioning module results in visual transitions that improve the capture of the desired video production style characteristics by an average of around 12% in comparison to other methods when measured with similarity metrics. We hope that our work serves as a foundation for exploring and understanding video production styles further.

31 July 2021 in arXiv
Verified
Preprint
Open access
PDF file
UBUttaran Bhattacharya
Uttaran Bhattacharya
NRNicholas Rewkowski
Nicholas Rewkowski
ABAbhishek Banerjee
Abhishek Banerjee
PGPooja Guhan
Pooja Guhan
ABAniket Bera
Aniket Bera
DMDinesh Manocha
Dinesh Manocha
We present Text2Gestures, a transformer-based learning method to interactively generate emotive full-body gestures for virtual agents aligned with natural language text inputs. Our method generates emotionally expressive gestures by utilizing the relevant biomechanical features for body expressions, also known as affective features. We also consider the intended task corresponding to the text and the target virtual agents' intended gender and handedness in our generation pipeline. We train and evaluate our network on the MPI Emotional Body Expressions Database and observe that our network produces state-of-the-art performance in generating gestures for virtual agents aligned with the text for narration or conversation. Our network can generate these gestures at interactive rates on a commodity GPU. We conduct a web-based user study and observe that around 91% of participants indicated our generated gestures to be at least plausible on a five-point Likert Scale. The emotions perceived by the participants from the gestures are also strongly positively correlated with the corresponding intended emotions, with a minimum Pearson coefficient of 0.77 in the valence dimension.

13 February 2023 in arXiv
Verified
Preprint
Open access
PDF file
PGPooja Guhan
Pooja Guhan
NANaman Awasthi
Naman Awasthi
KMKathryn McDonald
Kathryn McDonald
KBKristin Bussell
Kristin Bussell
DMDinesh Manocha
Dinesh Manocha
ABAniket Bera
Aniket Bera
We discuss MET, a learning-based algorithm proposed for perceiving a patient's level of engagement during telehealth sessions. We leverage latent vectors corresponding to Affective and Cognitive features frequently used in psychology literature to understand a person's level of engagement in a semi-supervised GAN-based framework. We showcase the efficacy of this method from the perspective of mental health and more specifically how this can be leveraged for a better understanding of patient engagement during telemental health sessions. To further the development of similar technologies that can be useful for telehealth, we also plan to release a dataset MEDICA containing 1299 video clips, each 3 seconds long and show experiments on the same. Our framework reports a 40% improvement in RMSE (Root Mean Squared Error) over state-of-the-art methods for engagement estimation. In our real-world tests, we also observed positive correlations between the working alliance inventory scores reported by psychotherapists. This indicates the potential of the proposed model to present patient engagement estimations that aligns well with the engagement measures used by psychotherapists.