Alex Stergiou's Projects
[T-IP 2023] Code for exponential adaptive pooling for PyTorch
Toolkit for downloading and processing Google's AudioSet dataset.
Scripts for download AudioSet
Implementation of "Slow-Fast Auditory Streams for Audio Recognition, ICASSP, 2021" in PyTorch
đ Use bootstrapping to normalise the data from opinion polls and make them similar to the population they represent.
[ICIP 2021] PyTorch code for "The Mind's Eye: Visualizing Class-Agnostic Features of CNNs" for generation of kernel features.
[ICCVW 2019] PyTorch code for Class Visualization Pyramid for intpreting spatio-temporal class-specific activations throughout the network
Script for creating SQL databases from video files to reduce random access overhead and inodes
Creating grid visualisations of dataset examples with the OpenCV library
Keras code and weights files for popular deep learning models.
Official PyTorch implementation of Dreaming to Distill: Data-free Knowledge Transfer via DeepInversion (CVPR 2020)
Cloned repo from raghakot/keras-vis
Applying Transfer Learning on Inception V3 model (weights trained on Imagenet) for the Oxford TV Human Interactions dataset. The network gets as inputs images extracted every 5 frames from videos.
Official repository for downloading and using Inter4K video interpolation dataset
Keras w/ Tensorflow backend implementation for 3D channel-wise convolutions
Official repository for downloading and using LAVIB
[ICCV 2023] Code implementation for "Leaping Into Memories: Space-Time Deep Feature Synthesis"
A sweet little collection of handy functions for video file downloading. đŧ
[ICASSP 2023] PyTorch code for "Play It Back: Iterative Attention for Audio Recognition"
[CVPR 2023] Code for action prediction from videos
Tensors and Dynamic neural networks in Python with strong GPU acceleration
[ICIP 2019] Implementation of Saliency Tubes for 3D Convolutions in Pytoch and Keras to localise the focus spatio-temporal regions of 3D CNNs.
[ICCV 2021] Code for approximated exponential maximum pooling
Code for : [Pattern Recognit. Lett. 2021] "Learn to cycle: Time-consistent feature discovery for action recognition" and [IJCNN 2021] "Multi-Temporal Convolutions for Human Action Recognition in Videos".
đšī¸ Small programs using TensorFlow and Keras as the main libraries to develop simple Neural Networks tasks
Class activation maps for your PyTorch models (CAM, Grad-CAM, Grad-CAM++, Smooth Grad-CAM++, Score-CAM, SS-CAM, IS-CAM)