TensorRF Explained: A milestone in Novel View SynthesisNovel View Synthesis has always been a prominent problem to address in 3D Computer Vision. Directly succeeding techniques like DeepVoxels and Scene Representation Networks, we have Neural Radiance Fields, or simply NeRFs. Neural Radiance Fields is a ...Dec 18, 2025·25 min read·1
Vision Transformers in JAXTransformers were introduced in 2017 and changed the course of Deep Learning research for many forthcoming years. Naturally, they were also adopted in the computer vision ecosystem which was till then, dominated by Convolutional Neural Networks or Co...Jun 21, 2025·15 min read·9
Getting Started with JAX: Modern Deep Learning FrameworkMachine Learning frameworks make our lives easier by offering built-in methods for faster prototyping and optimized GPU training. They also help by providing robust and scalable solutions which help engineers implement state-of-the-art models in a be...Jun 4, 2025·13 min read·6
Demystifying HOG Feature Extraction: A Beginner-Friendly GuideRemember how we used SVD to extract features from an image which helped our machine “see”? Today we are going to discuss one more such method. HOG (Histogram of Oriented Gradients) is a feature descriptor that captures the shape and structure of obje...Jun 1, 2025·10 min read·6
Guide to Implementing RANSAC in C++ ProgrammingMany real-world datasets are corrupted by noise and outliers. Traditional methods for model fitting, such as least squares, are highly susceptible to these outliers, as even a single outlier can significantly distort the estimated parameters. Techniq...May 29, 2025·11 min read·10
Camera Calibration with Zhang's MethodIn 3D Computer Vision, camera calibration is the process of determining specific camera parameters to complete operations with specified performance measures. These parameters include focal length, optical centre, skew coefficient, radian and tangent...May 24, 2025·10 min read·17
Backward Pass of CNNSince the forward pass ended on the Fully Connected Layer, it is only natural to begin the discussion of backpropagation from that layer. Fortunately for us, it is not any different than what we usually see in regular Neural Networks. Therefore, we c...Mar 16, 2025·11 min read·18
