Is Multiple Object Tracking a Matter of Specialization?

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Mancusi, Gianluca, Bernardi, Mattia, Panariello, Aniello, Porrello, Angelo, Cucchiara, Rita, Calderara, Simone
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910680667389952
author Mancusi, Gianluca
Bernardi, Mattia
Panariello, Aniello
Porrello, Angelo
Cucchiara, Rita
Calderara, Simone
author_facet Mancusi, Gianluca
Bernardi, Mattia
Panariello, Aniello
Porrello, Angelo
Cucchiara, Rita
Calderara, Simone
contents End-to-end transformer-based trackers have achieved remarkable performance on most human-related datasets. However, training these trackers in heterogeneous scenarios poses significant challenges, including negative interference - where the model learns conflicting scene-specific parameters - and limited domain generalization, which often necessitates expensive fine-tuning to adapt the models to new domains. In response to these challenges, we introduce Parameter-efficient Scenario-specific Tracking Architecture (PASTA), a novel framework that combines Parameter-Efficient Fine-Tuning (PEFT) and Modular Deep Learning (MDL). Specifically, we define key scenario attributes (e.g, camera-viewpoint, lighting condition) and train specialized PEFT modules for each attribute. These expert modules are combined in parameter space, enabling systematic generalization to new domains without increasing inference time. Extensive experiments on MOTSynth, along with zero-shot evaluations on MOT17 and PersonPath22 demonstrate that a neural tracker built from carefully selected modules surpasses its monolithic counterpart. We release models and code.
format Preprint
id arxiv_https___arxiv_org_abs_2411_00553
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Is Multiple Object Tracking a Matter of Specialization?
Mancusi, Gianluca
Bernardi, Mattia
Panariello, Aniello
Porrello, Angelo
Cucchiara, Rita
Calderara, Simone
Computer Vision and Pattern Recognition
Machine Learning
End-to-end transformer-based trackers have achieved remarkable performance on most human-related datasets. However, training these trackers in heterogeneous scenarios poses significant challenges, including negative interference - where the model learns conflicting scene-specific parameters - and limited domain generalization, which often necessitates expensive fine-tuning to adapt the models to new domains. In response to these challenges, we introduce Parameter-efficient Scenario-specific Tracking Architecture (PASTA), a novel framework that combines Parameter-Efficient Fine-Tuning (PEFT) and Modular Deep Learning (MDL). Specifically, we define key scenario attributes (e.g, camera-viewpoint, lighting condition) and train specialized PEFT modules for each attribute. These expert modules are combined in parameter space, enabling systematic generalization to new domains without increasing inference time. Extensive experiments on MOTSynth, along with zero-shot evaluations on MOT17 and PersonPath22 demonstrate that a neural tracker built from carefully selected modules surpasses its monolithic counterpart. We release models and code.
title Is Multiple Object Tracking a Matter of Specialization?
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2411.00553