ITI-CERTH participation in ActEV and AVS Tracks of TRECVID 2024
Creators
- Gkountakos, Konstantinos (Researcher)1
- Galanopoulos, Damianos (Researcher)1
- Leventakis, Antonios (Researcher)1
- Tsionkis, Georgios (Researcher)1
- Stavrothanasopoulos, Klearchos (Researcher)1
- Ioannidis, Konstantinos (Researcher)1
- Vrochidis, Stefanos (Researcher)1
- Mezaris, Vasileios (Researcher)1
- Kompatsiaris, Ioannis (Researcher)1
-
1.
Centre for Research and Technology Hellas
Description
This report presents the overview of the runs related to Ad-hoc Video Search (AVS) and Activities in Extended Video (ActEV) tasks on behalf of the ITI-CERTH team. Our participation in the AVS task involves a collection of five cross-modal deep network architectures and numerous pretrained models, which are used to calculate the similarities between video shots and queries. These calculated similarities serve as input to a trainable neural network that effectively combines them. During the retrieval stage, we also introduce a normalization step that utilizes both the current and previous AVS queries for revising the combined video shot-query similarities. For the ActEV task, we adapt our framework to support a rule-based classification to overcome the challenges of detecting and recognizing activities in a multi-label manner while experimenting with two separate activity classifiers.
Files
trecvid2024.pdf
Files
(861.8 kB)
| Name | Size | Download all |
|---|---|---|
|
md5:6f115fddb746e8adc42f8b2b1e7373e4
|
861.8 kB | Preview Download |