We propose a novel scene flow estimation approach to capture and infer 3D motions from point clouds.
Specifically, we distill a powerful commercial calibration tool in a recent neural network architecture on the large-scale SoccerNet dataset, composed of untrimmed broadcast videos of 500 soccer games.
In this paper, we focus our analysis on action spotting in soccer broadcast, which consists in temporally localizing the main actions in a soccer game.
Ranked #1 on Action Spotting on SoccerNet-v2
Since it is learnable, this mapping is allowed to be different per layer instead of being applied uniformly throughout the depth of the network.
MVTN exhibits clear performance gains in the tasks of 3D shape classification and 3D shape retrieval without the need for extra training supervision.
Ranked #1 on 3D Object Retrieval on ModelNet40
1 code implementation • 26 Nov 2020 • Adrien Deliège, Anthony Cioppa, Silvio Giancola, Meisam J. Seikavandi, Jacob V. Dueholm, Kamal Nasrollahi, Bernard Ghanem, Thomas B. Moeslund, Marc Van Droogenbroeck
In this work, we propose SoccerNet-v2, a novel large-scale corpus of manual annotations for the SoccerNet video dataset, along with open challenges to encourage more research in soccer understanding and broadcast production.
Ranked #1 on Camera shot segmentation on SoccerNet-v2
Extensive experiments show that using features trained with our novel pretraining strategy significantly improves the performance of recent state-of-the-art methods on three tasks: Temporal Action Localization, Action Proposal Generation, and Dense Video Captioning.
In this paper, we introduce a new NAS framework, dubbed LC-NAS, where we search for point cloud architectures that are constrained to a target latency.
We benchmark our loss on a large dataset of soccer videos, SoccerNet, and achieve an improvement of 12. 8% over the baseline.
Ranked #2 on Action Spotting on SoccerNet-v2
We integrate residual GCNs in a two-stage 3D object detection pipeline, where 3D object proposals are refined using a novel graph representation.
Ranked #12 on 3D Object Detection on KITTI Cars Hard
Successively, we refine our selection of 3D object candidates by exploiting the similarity capability of a 3D Siamese network.
A total of 6, 637 temporal annotations are automatically parsed from online match reports at a one minute resolution for three main classes of events (Goal, Yellow/Red Card, and Substitution).
Ranked #6 on Action Spotting on SoccerNet
In this work, we present TrackingNet, the first large-scale dataset and benchmark for object tracking in the wild.
We also present a technique to filter the pairs of 3D matched points based on the distribution of their distances.
In this work, we propose a method for three-dimensional (3D) reconstruction of wide crime scene, based on a Simultaneous Localization and Mapping (SLAM) approach.