Embedded Files
3D Shape Completion from Partial Point Clouds using a Deep Learning Model
3D Shape Completion from Partial Point Clouds using a Deep Learning Model
This is my MS research thesis. Estimating the 3D keypoints of an object has achieved considerable interest when the object is noisy or partially occluded. The purpose of this research is to confine the keypoints of an object extracted from a dense point cloud under the inspection of ground truth keypoints. We propose an approach for the generation of keypoints via point cloud completion.
This is my MS research thesis. Estimating the 3D keypoints of an object has achieved considerable interest when the object is noisy or partially occluded. The purpose of this research is to confine the keypoints of an object extracted from a dense point cloud under the inspection of ground truth keypoints. We propose an approach for the generation of keypoints via point cloud completion.
Comparison when pass partial PCD directly to keypoints estimator module or partial PCD to shape completion module by first complete its shape and then fed to the keypoints estimator module
Image Classification: A Unified Framework for Accurate Estimation using SIFT, SURF, and HOG Features with Random Forest, SVM, and ANN Classifiers"
Image Classification: A Unified Framework for Accurate Estimation using SIFT, SURF, and HOG Features with Random Forest, SVM, and ANN Classifiers"
This project presents a methodology for image classification encompassing Data Preprocessing, Feature Extraction (utilizing SIFT, SURF, and HOG methods), and Classification (employing Random Forest, Support Vector Machine, and Artificial Neural Network classifiers). The objective is to estimate classifier accuracy based on diverse feature extraction modules. Results spotlight the consistently exceptional performance of the Histogram of Oriented Gradients (HOG) method, achieving a flawless 97.68% accuracy across classifiers, underscoring its robust pattern recognition capabilities.
This project presents a methodology for image classification encompassing Data Preprocessing, Feature Extraction (utilizing SIFT, SURF, and HOG methods), and Classification (employing Random Forest, Support Vector Machine, and Artificial Neural Network classifiers). The objective is to estimate classifier accuracy based on diverse feature extraction modules. Results spotlight the consistently exceptional performance of the Histogram of Oriented Gradients (HOG) method, achieving a flawless 97.68% accuracy across classifiers, underscoring its robust pattern recognition capabilities.
Proposed methodology calculates the accuracy of three distinct classifiers by processing features through them
Power system load management and home automation using Arduino Mega(2560)
Power system load management and home automation using Arduino Mega(2560)
This is my final year project of Bachelor's. The project provided a solution for effective power management by assigning priorities to various loads, enabling their operation based on these priorities. When the power demand surpasses the specified threshold, lower-priority loads will gradually be deactivated.
This is my final year project of Bachelor's. The project provided a solution for effective power management by assigning priorities to various loads, enabling their operation based on these priorities. When the power demand surpasses the specified threshold, lower-priority loads will gradually be deactivated.
Page updated
Google Sites
Report abuse