REVIEW 1 cited by
2rd Place Solutions in the HC-STVG track of Person in Context Challenge 2021
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
2rd Place Solutions in the HC-STVG track of Person in Context Challenge 2021
read the original abstract
In this technical report, we present our solution to localize a spatio-temporal person in an untrimmed video based on a sentence. We achieve the second vIOU(0.30025) in the HC-STVG track of the 3rd Person in Context(PIC) Challenge. Our solution contains three parts: 1) human attributes information is extracted from the sentence, it is helpful to filter out tube proposals in the testing phase and supervise our classifier to learn appearance information in the training phase. 2) we detect humans with YoloV5 and track humans based on the DeepSort framework but replace the original ReID network with FastReID. 3) a visual transformer is used to extract cross-modal representations for localizing a spatio-temporal tube of the target person.
Forward citations
Cited by 1 Pith paper
-
ScanFocus: A Coarse-to-Fine Framework for Spatio-Temporal Video Grounding
A coarse-to-fine video-grounding framework improves temporal boundary accuracy by densely re-examining frames around coarse start/end predictions, achieving SOTA on HC-STVGv1/v2 and VidSTG.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.