Guglielmo Camporese

hands-segmentation-pytorch. A repo for training and finetuning models for hands segmentation.

cvaecaposr. Code for the Paper: "Conditional Variational Capsule Network for Open Set Recognition", Y. Guo, G. Camporese, W. Yang, A. Sperduti, L. Ballan, ICCV, 2021.

epic-kitchens-dataset-pytorch. Simple PyTorch Dataset for the EPIC-Kitchens-55 and EPIC-Kitchens-100 that handles frames and features (rgb, optical flow, and objects) for the Action Recognition and the Action Anticipation Tasks!

relvit. Official code of "Where are my Neighbors? Exploiting Patches Relations in Self-Supervised Vision Transformer", Guglielmo Camporese, Elena Izzo, Lamberto Ballan. BMVC, 2022.

learning_invariances_in_speech_recognition. In this work I investigate the speech command task developing and analyzing deep learning models. The state of the art technology uses convolutional neural networks (CNN) because of their intrinsic nature of learning correlated represen- tations as is the speech. In particular I develop different CNNs trained on the Google Speech Command Dataset and tested on different scenarios. A main problem on speech recognition consists in the differences on pronunciations of words among different people: one way of building an invariant model to variability is to augment the dataset perturbing the input. In this work I study two kind of augmentations: the Vocal Tract Length Perturbation (VTLP) and the Synchronous Overlap and Add (SOLA) that locally perturb the input in frequency and time respectively. The models trained on augmented data outperforms in accuracy, precision and recall all the models trained on the normal dataset. Also the design of CNNs has impact on learning invariances: the inception CNN architecture in fact helps on learning features that are invariant to speech variability using different kind of kernel sizes for convolution. Intuitively this is because of the implicit capability of the model on detecting different speech pattern lengths in the audio feature.

awesome-action-prediction-list. A curated list of papers and resources linked to action anticipation and early action recognition from videos.

audioscope. Mechanistic interpretability of Stable Audio 3 — extracting and steering a mood direction vector in the DiT residual stream.

useful. Useful scripts/commands/settings I mostly use for research/work.

math-unipd-booking-bot. Simple and easy to use python BOT for the COVID registration booking system of the math department @ unipd (torre archimede). This API creates an interface with the official website, with more useful functionalities.

visual-transformer-pytorch. An easy and minimal implementation of the Visual Transformer (ViT) in PyTorch, from scratch!

break_cifar10. Code for the Top-1 submission of contest of VCS AY 2020-2021, the Vision and Cognitive Service class, University of Padova, Italy.

Speech-Classification-a-dictionary-approach-with-MFCC-and-Dynamic-Time-Warping. Jupyter Notebook

glom. Minimal GLOM implementation in PyTorch.

deep-vector-quantization. What can we do with Vector Quantization on Deep Nets?