Srimat Chakradhar

Christoph Reich, Biplob Debnath, Deep Patel et al.

JPEG remains one of the most widespread lossy image coding methods. However, the non-differentiable nature of JPEG restricts the application in deep learning pipelines. Several differentiable approximations of JPEG have recently been proposed to address this issue. This paper conducts a comprehensive review of existing diff. JPEG approaches and identifies critical details that have been missed by previous methods. To this end, we propose a novel diff. JPEG approach, overcoming previous limitations. Our approach is differentiable w.r.t. the input image, the JPEG quality, the quantization tables, and the color conversion parameters. We evaluate the forward and backward performance of our diff. JPEG approach against existing methods. Additionally, extensive ablations are performed to evaluate crucial design choices. Our proposed diff. JPEG resembles the (non-diff.) reference implementation best, significantly surpassing the recent-best diff. approach by $3.47$dB (PSNR) on average. For strong compression rates, we can even improve PSNR by $9.51$dB. Strong adversarial attack results are yielded by our diff. JPEG, demonstrating the effective gradient approximation. Our code is available at https://github.com/necla-ml/Diff-JPEG.

Matin Mortaheb, Mohammad A. Amir Khojastepour, Srimat T. Chakradhar et al.

Deep learning based joint source-channel coding (JSCC) has demonstrated significant advancements in data reconstruction compared to separate source-channel coding (SSCC). This superiority arises from the suboptimality of SSCC when dealing with finite block-length data. Moreover, SSCC falls short in reconstructing data in a multi-user and/or multi-resolution fashion, as it only tries to satisfy the worst channel and/or the highest quality data. To overcome these limitations, we propose a novel deep learning multi-resolution JSCC framework inspired by the concept of multi-task learning (MTL). This proposed framework excels at encoding data for different resolutions through hierarchical layers and effectively decodes it by leveraging both current and past layers of encoded data. Moreover, this framework holds great potential for semantic communication, where the objective extends beyond data reconstruction to preserving specific semantic attributes throughout the communication process. These semantic features could be crucial elements such as class labels, essential for classification tasks, or other key attributes that require preservation. Within this framework, each level of encoded data can be carefully designed to retain specific data semantics. As a result, the precision of a semantic classifier can be progressively enhanced across successive layers, emphasizing the preservation of targeted semantics throughout the encoding and decoding stages. We conduct experiments on MNIST and CIFAR10 dataset. The experiment with both datasets illustrates that our proposed method is capable of surpassing the SSCC method in reconstructing data with different resolutions, enabling the extraction of semantic features with heightened confidence in successive layers. This capability is particularly advantageous for prioritizing and preserving more crucial semantic features within the datasets.

Matin Mortaheb, Mohammad A. Amir Khojastepour, Srimat T. Chakradhar et al.

Ensuring high-quality video content for wireless users has become increasingly vital. Nevertheless, maintaining a consistent level of video quality faces challenges due to the fluctuating encoded bitrate, primarily caused by dynamic video content, especially in live streaming scenarios. Video compression is typically employed to eliminate unnecessary redundancies within and between video frames, thereby reducing the required bandwidth for video transmission. The encoded bitrate and the quality of the compressed video depend on encoder parameters, specifically, the quantization parameter (QP). Poor choices of encoder parameters can result in reduced bandwidth efficiency and high likelihood of non-conformance. Non-conformance refers to the violation of the peak signal-to-noise ratio (PSNR) constraint for an encoded video segment. To address these issues, a real-time deep learning-based H.264 controller is proposed. This controller dynamically estimates the optimal encoder parameters based on the content of a video chunk with minimal delay. The objective is to maintain video quality in terms of PSNR above a specified threshold while minimizing the average bitrate of the compressed video. Experimental results, conducted on both QCIF dataset and a diverse range of random videos from public datasets, validate the effectiveness of this approach. Notably, it achieves improvements of up to 2.5 times in average bandwidth usage compared to the state-of-the-art adaptive bitrate video streaming, with a negligible non-conformance probability below $10^{-2}$.

Sibendu Paul, Kunal Rao, Giuseppe Coviello et al.

It is a common practice to think of a video as a sequence of images (frames), and re-use deep neural network models that are trained only on images for similar analytics tasks on videos. In this paper, we show that this leap of faith that deep learning models that work well on images will also work well on videos is actually flawed. We show that even when a video camera is viewing a scene that is not changing in any human-perceptible way, and we control for external factors like video compression and environment (lighting), the accuracy of video analytics application fluctuates noticeably. These fluctuations occur because successive frames produced by the video camera may look similar visually, but these frames are perceived quite differently by the video analytics applications. We observed that the root cause for these fluctuations is the dynamic camera parameter changes that a video camera automatically makes in order to capture and produce a visually pleasing video. The camera inadvertently acts as an unintentional adversary because these slight changes in the image pixel values in consecutive frames, as we show, have a noticeably adverse impact on the accuracy of insights from video analytics tasks that re-use image-trained deep learning models. To address this inadvertent adversarial effect from the camera, we explore the use of transfer learning techniques to improve learning in video analytics tasks through the transfer of knowledge from learning on image analytics tasks. In particular, we show that our newly trained Yolov5 model reduces fluctuation in object detection across frames, which leads to better tracking of objects(40% fewer mistakes in tracking). Our paper also provides new directions and techniques to mitigate the camera's adversarial effect on deep learning models used for video analytics applications.

Matin Mortaheb, Mohammad A. Amir Khojastepour, Srimat T. Chakradhar et al.

Providing wireless users with high-quality video content has become increasingly important. However, ensuring consistent video quality poses challenges due to variable encoded bitrate caused by dynamic video content and fluctuating channel bitrate caused by wireless fading effects. Suboptimal selection of encoder parameters can lead to video quality loss due to underutilized bandwidth or the introduction of video artifacts due to packet loss. To address this, a real-time deep learning based H.264 controller is proposed. This controller leverages instantaneous channel quality data driven from the physical layer, along with the video chunk, to dynamically estimate the optimal encoder parameters with a negligible delay in real-time. The objective is to maintain an encoded video bitrate slightly below the available channel bitrate. Experimental results, conducted on both QCIF dataset and a diverse selection of random videos from public datasets, validate the effectiveness of the approach. Remarkably, improvements of 10-20 dB in PSNR with repect to the state-of-the-art adaptive bitrate video streaming is achieved, with an average packet drop rate as low as 0.002.

Sibendu Paul, Kunal Rao, Giuseppe Coviello et al.

IoT sensors, especially video cameras, are ubiquitously deployed around the world to perform a variety of computer vision tasks in several verticals including retail, healthcare, safety and security, transportation, manufacturing, etc. To amortize their high deployment effort and cost, it is desirable to perform multiple video analytics tasks, which we refer to as Analytical Units (AUs), off the video feed coming out of every camera. In this paper, we first show that in a multi-AU setting, changing the camera setting has disproportionate impact on different AUs performance. In particular, the optimal setting for one AU may severely degrade the performance for another AU, and further the impact on different AUs varies as the environmental condition changes. We then present Elixir, a system to enhance the video stream quality for multiple analytics on a video stream. Elixir leverages Multi-Objective Reinforcement Learning (MORL), where the RL agent caters to the objectives from different AUs and adjusts the camera setting to simultaneously enhance the performance of all AUs. To define the multiple objectives in MORL, we develop new AU-specific quality estimator values for each individual AU. We evaluate Elixir through real-world experiments on a testbed with three cameras deployed next to each other (overlooking a large enterprise parking lot) running Elixir and two baseline approaches, respectively. Elixir correctly detects 7.1% (22,068) and 5.0% (15,731) more cars, 94% (551) and 72% (478) more faces, and 670.4% (4975) and 158.6% (3507) more persons than the default-setting and time-sharing approaches, respectively. It also detects 115 license plates, far more than the time-sharing approach (7) and the default setting (0).

Md Adnan Arefeen, Biplob Debnath, Ravi K. Rajendran et al.

In satellite applications, user queries often take the form of open-ended natural language, extending beyond a fixed set of predefined categories. This open-vocabulary nature poses significant challenges for retrieving relevant image tiles, as the retrieval system must generalize to a wide range of unseen objects and concepts. While vision-language models (VLMs) such as CLIP are widely used for text-image retrieval, even fine-tuned variants often struggle to accurately align such queries with satellite imagery. To address this, we propose Open-SAT, a training-free query embedding refinement algorithm that operates at inference time to improve alignment between user queries and satellite image content. Open-SAT uses VLMs to compute embeddings for image tiles, which are stored in a vector database for efficient retrieval. At query time, it leverages Large Language Models (LLMs) to refine the text embeddings by incorporating contextual information about objects of interest and their surroundings. A threshold-free retrieval mechanism further enhances accuracy and efficiency. Experimental results in three public benchmarks demonstrate that Open-SAT improves the F1 score by up to 16.04%, while retrieving a comparable number of image tiles. These results demonstrate the effectiveness of Open-SAT in open-vocabulary satellite image retrieval, leveraging LLM guidance without the need for additional training or supervision.

Sarosij Bose, Ravi K. Rajendran, Biplob Debnath et al.

Radiology Report Generation (RRG) is a critical step toward automating healthcare workflows, facilitating accurate patient assessments, and reducing the workload of medical professionals. Despite recent progress in Large Medical Vision-Language Models (Med-VLMs), generating radiology reports that are both visually grounded and clinically accurate remains a significant challenge. Existing approaches often rely on large labeled corpora for pre-training, costly task-specific preference data, or retrieval-based methods. However, these strategies do not adequately mitigate hallucinations arising from poor cross-modal alignment between visual and linguistic representations. To address these limitations, we propose VALOR:Visual Alignment of Medical Vision-Language Models for GrOunded Radiology Report Generation. Our method introduces a reinforcement learning-based post-alignment framework utilizing Group-Relative Proximal Optimization (GRPO). The training proceeds in two stages: (1) improving the Med-VLM with textual rewards to encourage clinically precise terminology, and (2) aligning the vision projection module of the textually grounded model with disease findings, thereby guiding attention toward image re gions most relevant to the diagnostic task. Extensive experiments on multiple benchmarks demonstrate that VALOR substantially improves factual accuracy and visual grounding, achieving significant performance gains over state-of-the-art report generation methods.

Sibendu Paul, Kunal Rao, Giuseppe Coviello et al.

Cameras are increasingly being deployed in cities, enterprises and roads world-wide to enable many applications in public safety, intelligent transportation, retail, healthcare and manufacturing. Often, after initial deployment of the cameras, the environmental conditions and the scenes around these cameras change, and our experiments show that these changes can adversely impact the accuracy of insights from video analytics. This is because the camera parameter settings, though optimal at deployment time, are not the best settings for good-quality video capture as the environmental conditions and scenes around a camera change during operation. Capturing poor-quality video adversely affects the accuracy of analytics. To mitigate the loss in accuracy of insights, we propose a novel, reinforcement-learning based system APT that dynamically, and remotely (over 5G networks), tunes the camera parameters, to ensure a high-quality video capture, which mitigates any loss in accuracy of video analytics. As a result, such tuning restores the accuracy of insights when environmental conditions or scene content change. APT uses reinforcement learning, with no-reference perceptual quality estimation as the reward function. We conducted extensive real-world experiments, where we simultaneously deployed two cameras side-by-side overlooking an enterprise parking lot (one camera only has manufacturer-suggested default setting, while the other camera is dynamically tuned by APT during operation). Our experiments demonstrated that due to dynamic tuning by APT, the analytics insights are consistently better at all times of the day: the accuracy of object detection video analytics application was improved on average by ~ 42%. Since our reward function is independent of any analytics task, APT can be readily used for different video analytics tasks.

Md Adnan Arefeen, Biplob Debnath, Srimat Chakradhar

Question-answering (QA) is a significant application of Large Language Models (LLMs), shaping chatbot capabilities across healthcare, education, and customer service. However, widespread LLM integration presents a challenge for small businesses due to the high expenses of LLM API usage. Costs rise rapidly when domain-specific data (context) is used alongside queries for accurate domain-specific LLM responses. One option is to summarize the context by using LLMs and reduce the context. However, this can also filter out useful information that is necessary to answer some domain-specific queries. In this paper, we shift from human-oriented summarizers to AI model-friendly summaries. Our approach, LeanContext, efficiently extracts $k$ key sentences from the context that are closely aligned with the query. The choice of $k$ is neither static nor random; we introduce a reinforcement learning technique that dynamically determines $k$ based on the query and context. The rest of the less important sentences are reduced using a free open source text reduction method. We evaluate LeanContext against several recent query-aware and query-unaware context reduction approaches on prominent datasets (arxiv papers and BBC news articles). Despite cost reductions of $37.29\%$ to $67.81\%$, LeanContext's ROUGE-1 score decreases only by $1.41\%$ to $2.65\%$ compared to a baseline that retains the entire context (no summarization). Additionally, if free pretrained LLM-based summarizers are used to reduce context (into human consumable summaries), LeanContext can further modify the reduced context to enhance the accuracy (ROUGE-1 score) by $13.22\%$ to $24.61\%$.

Md Adnan Arefeen, Biplob Debnath, Md Yusuf Sarwar Uddin et al.

Retrieval-augmented generation (RAG) systems combine the strengths of language generation and information retrieval to power many real-world applications like chatbots. Use of RAG for understanding of videos is appealing but there are two critical limitations. One-time, upfront conversion of all content in large corpus of videos into text descriptions entails high processing times. Also, not all information in the rich video data is typically captured in the text descriptions. Since user queries are not known apriori, developing a system for video to text conversion and interactive querying of video data is challenging. To address these limitations, we propose an incremental RAG system called iRAG, which augments RAG with a novel incremental workflow to enable interactive querying of a large corpus of videos. Unlike traditional RAG, iRAG quickly indexes large repositories of videos, and in the incremental workflow, it uses the index to opportunistically extract more details from select portions of the videos to retrieve context relevant to an interactive user query. Such an incremental workflow avoids long video to text conversion times, and overcomes information loss issues due to conversion of video to text, by doing on-demand query-specific extraction of details in video data. This ensures high quality of responses to interactive user queries that are often not known apriori. To the best of our knowledge, iRAG is the first system to augment RAG with an incremental workflow to support efficient interactive querying of a large corpus of videos. Experimental results on real-world datasets demonstrate 23x to 25x faster video to text ingestion, while ensuring that latency and quality of responses to interactive user queries is comparable to responses from a traditional RAG where all video data is converted to text upfront before any user querying.

Matin Mortaheb, Mohammad A. Amir Khojastepour, Srimat T. Chakradhar et al.

Retrieval-augmented generation (RAG) enhances large language models (LLMs) by incorporating external knowledge to generate a response within a context with improved accuracy and reduced hallucinations. However, multi-modal RAG systems face unique challenges: (i) the retrieval process may select irrelevant entries to user query (e.g., images, documents), and (ii) vision-language models or multi-modal language models like GPT-4o may hallucinate when processing these entries to generate RAG output. In this paper, we aim to address the first challenge, i.e, improving the selection of relevant context from the knowledge-base in retrieval phase of the multi-modal RAG. Specifically, we leverage the relevancy score (RS) measure designed in our previous work for evaluating the RAG performance to select more relevant entries in retrieval process. The retrieval based on embeddings, say CLIP-based embedding, and cosine similarity usually perform poorly particularly for multi-modal data. We show that by using a more advanced relevancy measure, one can enhance the retrieval process by selecting more relevant pieces from the knowledge-base and eliminate the irrelevant pieces from the context by adaptively selecting up-to-$k$ entries instead of fixed number of entries. Our evaluation using COCO dataset demonstrates significant enhancement in selecting relevant context and accuracy of the generated response.

Murugan Sankaradas, Ravi K. Rajendran, Srimat T. Chakradhar

Extracting real-time insights from multi-modal data streams from various domains such as healthcare, intelligent transportation, and satellite remote sensing remains a challenge. High computational demands and limited knowledge scope restrict the applicability of Multi-Modal Large Language Models (MM-LLMs) on these data streams. Traditional Retrieval-Augmented Generation (RAG) systems address knowledge limitations of these models, but suffer from slow preprocessing, making them unsuitable for real-time analysis. We propose StreamingRAG, a novel RAG framework designed for streaming data. StreamingRAG constructs evolving knowledge graphs capturing scene-object-entity relationships in real-time. The knowledge graph achieves temporal-aware scene representations using MM-LLMs and enables timely responses for specific events or user queries. StreamingRAG addresses limitations in existing methods, achieving significant improvements in real-time analysis (5-6x faster throughput), contextual accuracy (through a temporal knowledge graph), and reduced resource consumption (using lightweight models by 2-3x).

Matin Mortaheb, Mohammad A. Amir Khojastepour, Srimat T. Chakradhar et al.

Retrieval-augmented generation (RAG) improves large language models (LLMs) by using external knowledge to guide response generation, reducing hallucinations. However, RAG, particularly multi-modal RAG, can introduce new hallucination sources: (i) the retrieval process may select irrelevant pieces (e.g., documents, images) as raw context from the database, and (ii) retrieved images are processed into text-based context via vision-language models (VLMs) or directly used by multi-modal language models (MLLMs) like GPT-4o, which may hallucinate. To address this, we propose a novel framework to evaluate the reliability of multi-modal RAG using two performance measures: (i) the relevancy score (RS), assessing the relevance of retrieved entries to the query, and (ii) the correctness score (CS), evaluating the accuracy of the generated response. We train RS and CS models using a ChatGPT-derived database and human evaluator samples. Results show that both models achieve ~88% accuracy on test data. Additionally, we construct a 5000-sample human-annotated database evaluating the relevancy of retrieved pieces and the correctness of response statements. Our RS model aligns with human preferences 20% more often than CLIP in retrieval, and our CS model matches human preferences ~91% of the time. Finally, we assess various RAG systems' selection and generation performances using RS and CS.

Murugan Sankaradas, Kunal Rao, Ravi Rajendran et al.

Edge computing and 5G have made it possible to perform analytics closer to the source of data and achieve super-low latency response times, which is not possible with centralized cloud deployment. In this paper, we present a novel fever-screening system, which uses edge machine learning techniques and leverages private 5G to accurately identify and screen individuals with fever in real-time. Particularly, we present deep-learning based novel techniques for fusion and alignment of cross-spectral visual and thermal data streams at the edge. Our novel Cross-Spectral Generative Adversarial Network (CS-GAN) synthesizes visual images that have the key, representative object level features required to uniquely associate objects across visual and thermal spectrum. Two key features of CS-GAN are a novel, feature-preserving loss function that results in high-quality pairing of corresponding cross-spectral objects, and dual bottleneck residual layers with skip connections (a new, network enhancement) to not only accelerate real-time inference, but to also speed up convergence during model training at the edge. To the best of our knowledge, this is the first technique that leverages 5G networks and limited edge resources to enable real-time feature-level association of objects in visual and thermal streams (30 ms per full HD frame on an Intel Core i7-8650 4-core, 1.9GHz mobile processor). To the best of our knowledge, this is also the first system to achieve real-time operation, which has enabled fever screening of employees and guests in arenas, theme parks, airports and other critical facilities. By leveraging edge computing and 5G, our fever screening system is able to achieve 98.5% accuracy and is able to process about 5X more people when compared to a centralized cloud deployment.

Sibendu Paul, Kunal Rao, Giuseppe Coviello et al.

In Video Analytics Pipelines (VAP), Analytics Units (AUs) such as object detection and face recognition running on remote servers critically rely on surveillance cameras to capture high-quality video streams in order to achieve high accuracy. Modern IP cameras come with a large number of camera parameters that directly affect the quality of the video stream capture. While a few of such parameters, e.g., exposure, focus, white balance are automatically adjusted by the camera internally, the remaining ones are not. We denote such camera parameters as non-automated (NAUTO) parameters. In this paper, we first show that environmental condition changes can have significant adverse effect on the accuracy of insights from the AUs, but such adverse impact can potentially be mitigated by dynamically adjusting NAUTO camera parameters in response to changes in environmental conditions. We then present CamTuner, to our knowledge, the first framework that dynamically adapts NAUTO camera parameters to optimize the accuracy of AUs in a VAP in response to adverse changes in environmental conditions. CamTuner is based on SARSA reinforcement learning and it incorporates two novel components: a light-weight analytics quality estimator and a virtual camera that drastically speed up offline RL training. Our controlled experiments and real-world VAP deployment show that compared to a VAP using the default camera setting, CamTuner enhances VAP accuracy by detecting 15.9% additional persons and 2.6%-4.2% additional cars (without any false positives) in a large enterprise parking lot and 9.7% additional cars in a 5G smart traffic intersection scenario, which enables a new usecase of accurate and reliable automatic vehicle collision prediction (AVCP). CamTuner opens doors for new ways to significantly enhance video analytics accuracy beyond incremental improvements from refining deep-learning models.

Sibendu Paul, Utsav Drolia, Y. Charlie Hu et al.

Millions of cameras at edge are being deployed to power a variety of different deep learning applications. However, the frames captured by these cameras are not always pristine - they can be distorted due to lighting issues, sensor noise, compression etc. Such distortions not only deteriorate visual quality, they impact the accuracy of deep learning applications that process such video streams. In this work, we introduce AQuA, to protect application accuracy against such distorted frames by scoring the level of distortion in the frames. It takes into account the analytical quality of frames, not the visual quality, by learning a novel metric, classifier opinion score, and uses a lightweight, CNN-based, object-independent feature extractor. AQuA accurately scores distortion levels of frames and generalizes to multiple different deep learning applications. When used for filtering poor quality frames at edge, it reduces high-confidence errors for analytics applications by 17%. Through filtering, and due to its low overhead (14ms), AQuA can also reduce computation time and average bandwidth usage by 25%.

Linnan Wang, Yi Yang, Martin Renqiang Min et al.

SGD is the widely adopted method to train CNN. Conceptually it approximates the population with a randomly sampled batch; then it evenly trains batches by conducting a gradient update on every batch in an epoch. In this paper, we demonstrate Sampling Bias, Intrinsic Image Difference and Fixed Cycle Pseudo Random Sampling differentiate batches in training, which then affect learning speeds on them. Because of this, the unbiased treatment of batches involved in SGD creates improper load balancing. To address this issue, we present Inconsistent Stochastic Gradient Descent (ISGD) to dynamically vary training effort according to learning statuses on batches. Specifically ISGD leverages techniques in Statistical Process Control to identify a undertrained batch. Once a batch is undertrained, ISGD solves a new subproblem, a chasing logic plus a conservative constraint, to accelerate the training on the batch while avoid drastic parameter changes. Extensive experiments on a variety of datasets demonstrate ISGD converges faster than SGD. In training AlexNet, ISGD is 21.05\% faster than SGD to reach 56\% top1 accuracy under the exactly same experiment setup. We also extend ISGD to work on multiGPU or heterogeneous distributed system based on data parallelism, enabling the batch size to be the key to scalability. Then we present the study of ISGD batch size to the learning rate, parallelism, synchronization cost, system saturation and scalability. We conclude the optimal ISGD batch size is machine dependent. Various experiments on a multiGPU system validate our claim. In particular, ISGD trains AlexNet to 56.3% top1 and 80.1% top5 accuracy in 11.5 hours with 4 NVIDIA TITAN X at the batch size of 1536.