gautamHCSCV

AI-ML

Anomaly-Detection

Due to the intractability of characterizing everything that looks unlike the normal data, anomaly detection (AD) is traditionally treated as an unsupervised problem utilizing only normal samples. However, it has recently been found that unsupervised image AD can be drastically improved.

Attacks-and-Defence

Adversarial attacks, their identifications and defenses against them

Bias-Mitigation-using-Knowledge-Distillation

A teacher model with minimal bias or calibrate its outputs to reduce bias transfer during distillation. Focus on distilling information from teacher features that are less prone to bias, like semantic representations instead of raw activations.

Chest-Image-classification

We design a lightweight CNN architecture for the chest x-ray classi-32 fication task by introducing ExLNet which uses a novel DCISE blocks to reduce the33 computational burden. We show the effectiveness of the proposed architecture through34 various experiments performed on publicly available datasets.

Image-Anonymization-using-Adversarial-Attacks

The Fast Gradient Sign Method (FGSM) combines a white box approach with a misclassification goal. It tricks a neural network model into making wrong predictions. We use this technique to anonymize images.

Modelling-of-Inhomogeneous-Objects

Modelling viscoelastic deformable objects refers to the process of representing and simulating the behavior of materials that exhibit both viscous and elastic properties when subjected to external forces or deformations.

Modelling_Viscoelastic_Objects

This paper proposes an alternative data-driven hap- tic modeling method of homogeneous deformable objects based on a CatBoost approach – a variant of gradient boosting machine learning approach. In this approach, decision trees are trained sequentially to learn the required mapping function for modeling the objects.

Number_Plate_detection-using-YOLO-v7

YOLOv7 is the new state-of-the-art object detector in the YOLO family. According to the paper, it is the fastest and most accurate real-time object detector to date. According to the YOLOv7 paper, the best model scored 56.8% Average Precision (AP), which is the highest among all known object detectors.

Transformers-Applications

A transformer is a deep learning model. It is distinguished by its adoption of self-attention, differentially weighting the significance of each part of the input (which includes the recursive output) data. It is used primarily in the fields of natural language processing (NLP) and computer vision (CV).

gautamHCSCV/Chest-Image-classification

We design a lightweight CNN architecture for the chest x-ray classi-32 fication task by introducing ExLNet which uses a novel DCISE blocks to reduce the33 computational burden. We show the effectiveness of the proposed architecture through34 various experiments performed on publicly available datasets.

gautamHCSCV/Modelling-of-Inhomogeneous-Objects

Modelling viscoelastic deformable objects refers to the process of representing and simulating the behavior of materials that exhibit both viscous and elastic properties when subjected to external forces or deformations.

gautamHCSCV/Modelling_Viscoelastic_Objects

This paper proposes an alternative data-driven hap- tic modeling method of homogeneous deformable objects based on a CatBoost approach – a variant of gradient boosting machine learning approach. In this approach, decision trees are trained sequentially to learn the required mapping function for modeling the objects.

gautamHCSCV/Number_Plate_detection-using-YOLO-v7

YOLOv7 is the new state-of-the-art object detector in the YOLO family. According to the paper, it is the fastest and most accurate real-time object detector to date. According to the YOLOv7 paper, the best model scored 56.8% Average Precision (AP), which is the highest among all known object detectors.

gautamHCSCV/AI-ML

gautamHCSCV/Anomaly-Detection

Due to the intractability of characterizing everything that looks unlike the normal data, anomaly detection (AD) is traditionally treated as an unsupervised problem utilizing only normal samples. However, it has recently been found that unsupervised image AD can be drastically improved.

gautamHCSCV/Attacks-and-Defence

Adversarial attacks, their identifications and defenses against them

gautamHCSCV/Bias-Mitigation-using-Knowledge-Distillation

A teacher model with minimal bias or calibrate its outputs to reduce bias transfer during distillation. Focus on distilling information from teacher features that are less prone to bias, like semantic representations instead of raw activations.

gautamHCSCV/Chest-Radiograph-Anomaly-Detection

gautamHCSCV/Federated-Learning-Methods-Comparison

We utilize the Adversarial Model Perturbations (AMP) regularizer to regularize clients’ models. The AMP regulzaizer is based on perturbing the model parameters so as to get a more generalized model. The claim of AMP regularizer is to reach flat minima and therefore is expected to reach flat minima in FL settings as well.

gautamHCSCV/HFSM-Python

Hierarchical state machines are finite state machines whose states themselves can be other state machines. Hierarchy is a useful construct in many modeling formalisms and tools for software design, requirements and testing. We summarize recent work on hierarchical state machines with or without concurrency.

gautamHCSCV/Image-Anonymization-using-Adversarial-Attacks

The Fast Gradient Sign Method (FGSM) combines a white box approach with a misclassification goal. It tricks a neural network model into making wrong predictions. We use this technique to anonymize images.

gautamHCSCV/Transformers-Applications

A transformer is a deep learning model. It is distinguished by its adoption of self-attention, differentially weighting the significance of each part of the input (which includes the recursive output) data. It is used primarily in the fields of natural language processing (NLP) and computer vision (CV).

gautamHCSCV/Cell-Detection

Object Detection project

gautamHCSCV/Classification-Loss-Comparison

Densenet121 models is compared with Triple loss, Center loss and Cross-entropy loss. Cross-Entropy Loss Function. Also called logarithmic loss, log loss or logistic loss. Each predicted class probability is compared to the actual class desired output 0 or 1 and a score/loss is calculated that penalizes the probability based on how far it is from the actual expected value. Triplet loss is a loss function for machine learning algorithms where a reference input (called anchor) is compared to a matching input (called positive) and a non-matching input (called negative). Center loss reduces the distance of each data point to its class center. It is not as difficult to train as triplet loss and performance is not based on the selection process of the training data points(triplets). Combining it with a softmax loss, prevents embeddings from collapsing.

gautamHCSCV/CNN_Algos_Comparison

Comparative Study between different CNN algorithms and comparing them using various evaluation metics.

gautamHCSCV/Cow-Face-Recognition

Facial recognition is the process of identifying or verifying a person's identity using their face. It captures, analyzes, and compares patterns based on the person's facial details.

gautamHCSCV/Detecting-Emerging-Technologies-and-their-Evolution-using-Deep-Learning-and-Weak-Signal-Analysis-

We present a multi-layer quantitative approach able to identify future signs from scientific publications on hypersonics by leveraging deep learning and weak signal analysis. The proposed framework can help strategic planners and domain experts better identify and monitor emerging technology trends.

gautamHCSCV/Diffusion-For-Motion-Blur-Removal

gautamHCSCV/HIndi-OCR

Storage for Hindi OCR work

gautamHCSCV/Image-Classification-on-Edges-using-BNN

Classification of images based on edges only. The images are converted to their gray-scale format followed by application of sobel filter to detect the edges. Models are applied on this new set of images with only edges to detect their class.

gautamHCSCV/Image_Recolor_GANs

Grayscale to LAB format and grayscale to RGB format.

gautamHCSCV/Machine-Learning-Labs

gautamHCSCV/Telemedicine_chatbot

gautamHCSCV/UnityTutorials-FiniteStateMachines

The code for a short tutorial on finite state machines and how to use them for basic 2D physics-based player movement in Unity/C# (in text or video format).

gautamHCSCV/Unsupervised-Detections

gautamHCSCV/Visual-Haptic-Generation

Studies that have been done in the field of psychophysics so far have shown that humans frequently engage in cross-modal visual-tactile perception when they go about their daily lives. To build an algorithmic mapping from one modality space to another.

gautamHCSCV/Wechat-QR-reader-Dockerization

Docker images are immutable, which means they can't be modified once created. If changes need to be made to an image, you must create a new image with the desired modifications. In contrast, containers are mutable and allow modifications during runtime.

gautamHCSCV/yolov7_traffic_detecitons

Yolo-v7 model for helmet detection.

gautamHCSCV/yolov8-pipeline