Inspired by YOLO, using pre-trained weights and image splitting, Here I am localizing and classifying the objects to understand in-depth how it works.
-
Updated
Feb 14, 2021 - Jupyter Notebook
Inspired by YOLO, using pre-trained weights and image splitting, Here I am localizing and classifying the objects to understand in-depth how it works.
Resent Client
Python script leveraging pre-trained ResNet18 for extracting video features from the YouTube Dataset, enabling LSTM-based action recognition models.
4-stage CV + ML pipeline for real-time driver drowsiness detection · ROC-AUC 0.964 · F1 0.923 · PyTorch · RetinaFace · ResNet-18 · Random Forest
Nextjs starter kit
Developed an image colorization pipeline utilizing Non-Local Means, Total Variation, and Wavelet Denoising for noise removal, with SIFT and ResNet backbones for feature extraction on ImageNet-1k, and enhanced spatial consistency through joint bilateral filtering and 4K upscaling via bilinear interpolation.
A modern pytorch implementation of CVPR 2016 paper: "A Hierarchical Deep Temporal Model for Group Activity Recognition."
To associate your repository with the resent topic, visit your repo's landing page and select "manage topics."