Refining Action Boundaries for One-stage Detection

Hanyuan Wang; Majid Mirmehdi; Dima  Damen; Toby J Perrett

Refining Action Boundaries for One-stage Detection

Hanyuan Wang, Majid Mirmehdi, Dima Damen, Toby J Perrett

Research output: Contribution to conference › Conference Paper › peer-review

Abstract

Current one-stage action detection methods, which simultaneously predict action boundaries and the corresponding class, do not estimate or use a measure of confidence in their boundary predictions, which can lead to inaccurate boundaries. We incorporate the estimation of boundary confidence into one-stage anchor-free detection, through an additional prediction head that predicts the refined boundaries with higher confidence. We obtain state-of-the-art performance on the challenging EPIC-KITCHENS-100 action detection as well as the standard THUMOS14 action detection benchmarks, and achieve improvement on the ActivityNet-1.3 benchmark.

Original language	English
Number of pages	8
Publication status	Published - 25 Oct 2022
Event	The 18th IEEE Int. Conf. on Advanced Video and Signal-Based Surveillance - Duration: 29 Nov 2022 → 2 Dec 2022

Conference

Conference	The 18th IEEE Int. Conf. on Advanced Video and Signal-Based Surveillance
Period	29/11/22 → 2/12/22

Access to Document

https://arxiv.org/abs/2210.14284

Handle.net

Persistent link

HPC (High Performance Computing) and HTC (High Throughput Computing) Facilities
Alam, S. R. (Manager), Williams, D. A. G. (Manager), Eccleston, P. E. (Manager) & Greene, D. (Manager)
Facility/equipment: Facility

Cite this

@conference{d695352424564b509fcd05caaf498eeb,

title = "Refining Action Boundaries for One-stage Detection",

abstract = "Current one-stage action detection methods, which simultaneously predict action boundaries and the corresponding class, do not estimate or use a measure of confidence in their boundary predictions, which can lead to inaccurate boundaries. We incorporate the estimation of boundary confidence into one-stage anchor-free detection, through an additional prediction head that predicts the refined boundaries with higher confidence. We obtain state-of-the-art performance on the challenging EPIC-KITCHENS-100 action detection as well as the standard THUMOS14 action detection benchmarks, and achieve improvement on the ActivityNet-1.3 benchmark.",

author = "Hanyuan Wang and Majid Mirmehdi and Dima Damen and Perrett, {Toby J}",

year = "2022",

month = oct,

day = "25",

language = "English",

note = "The 18th IEEE Int. Conf. on Advanced Video and Signal-Based Surveillance ; Conference date: 29-11-2022 Through 02-12-2022",

}

TY - CONF

T1 - Refining Action Boundaries for One-stage Detection

AU - Wang, Hanyuan

AU - Mirmehdi, Majid

AU - Damen, Dima

AU - Perrett, Toby J

PY - 2022/10/25

Y1 - 2022/10/25

N2 - Current one-stage action detection methods, which simultaneously predict action boundaries and the corresponding class, do not estimate or use a measure of confidence in their boundary predictions, which can lead to inaccurate boundaries. We incorporate the estimation of boundary confidence into one-stage anchor-free detection, through an additional prediction head that predicts the refined boundaries with higher confidence. We obtain state-of-the-art performance on the challenging EPIC-KITCHENS-100 action detection as well as the standard THUMOS14 action detection benchmarks, and achieve improvement on the ActivityNet-1.3 benchmark.

AB - Current one-stage action detection methods, which simultaneously predict action boundaries and the corresponding class, do not estimate or use a measure of confidence in their boundary predictions, which can lead to inaccurate boundaries. We incorporate the estimation of boundary confidence into one-stage anchor-free detection, through an additional prediction head that predicts the refined boundaries with higher confidence. We obtain state-of-the-art performance on the challenging EPIC-KITCHENS-100 action detection as well as the standard THUMOS14 action detection benchmarks, and achieve improvement on the ActivityNet-1.3 benchmark.

M3 - Conference Paper

T2 - The 18th IEEE Int. Conf. on Advanced Video and Signal-Based Surveillance

Y2 - 29 November 2022 through 2 December 2022

ER -

Refining Action Boundaries for One-stage Detection

Abstract

Conference

Access to Document

Handle.net

Fingerprint

Equipment

HPC (High Performance Computing) and HTC (High Throughput Computing) Facilities

Cite this