Smart Director: An Event-Driven Directing System for Live Broadcasting
Not on the map. Why?
Abstract (arXiv)
Live video broadcasting normally requires a multitude of skills and expertise with domain knowledge to enable multi-camera productions. As the number of cameras keep increasing, directing a live sports broadcast has now become more complicated and challenging than ever before. The broadcast directors need to be much more concentrated, responsive, and knowledgeable, during the production. To relieve the directors from their intensive efforts, we develop an innovative automated sports broadcast directing system, called Smart Director, which aims at mimicking the typical human-in-the-loop broadcasting process to automatically create near-professional broadcasting programs in real-time by using a set of advanced multi-view video analysis algorithms. Inspired by the so-called "three-event" construction of sports broadcast, we build our system with an event-driven pipeline consisting of three consecutive novel components: 1) the Multi-view Event Localization to detect events by modeling multi-view correlations, 2) the Multi-view Highlight Detection to rank camera views by the visual importance for view selection, 3) the Auto-Broadcasting Scheduler to control the production of broadcasting videos. To our best knowledge, our system is the first end-to-end automated directing system for multi-camera sports broadcasting, completely driven by the semantic understanding of sports events. It is also the first system to solve the novel problem of multi-view joint event detection by cross-view relation modeling. We conduct both objective and subjective evaluations on a real-world multi-camera soccer dataset, which demonstrate the quality of our auto-generated videos is comparable to that of the human-directed. Thanks to its faster response, our system is able to capture more fast-passing and short-duration events which are usually missed by human directors.
Key facts. Cited by 7 papers in the map (data through 7 Oct 2026). Builds on 26 of its 58 references that are in the map. First posted on arXiv 11 Jan 2022.
Builds on
26 of 58 references.
References · 26
- 2016Image-to-Image Translation with Conditional Adversarial Networkscited by 4,510
- 2017Learning Spatio-Temporal Representation with Pseudo-3D Residual Networkscited by 386
- 2017Temporal Action Detection with Structured Segment Networkscited by 265
- 2018Exploring Visual Relationship for Image Captioningcited by 215
- 2020X-Linear Attention Networks for Image Captioningcited by 198
- 2016Video Summarization with Long Short-term Memorycited by 142
- 2019Exploring Object Relation in Mean Teacher for Cross-Domain Detectioncited by 136
- 2019Gaussian Temporal Awareness Networks for Action Localizationcited by 135
- 2017Single Shot Temporal Action Detectioncited by 123
- 2019Relation Distillation Networks for Video Object Detectioncited by 77
- 2020SeCo: Exploring Sequence Supervision for Unsupervised Representation Learningcited by 65
- 2018Jointly Localizing and Describing Events for Dense Video Captioningcited by 60
14 more
- 2019Less is More: Learning Highlight Detection from Video Durationcited by 60
- 2018CTAP: Complementary Temporal Action Proposal Generationcited by 46
- 2017FaceBoxes: A CPU Real-time Face Detector with High Accuracycited by 44
- 2019Hierarchy Parsing for Image Captioningcited by 44
- 2019Mocycle-GAN: Unpaired Video-to-Video Translationcited by 37
- 2019Dance with Flow: Two-in-One Stream Action Detectioncited by 30
- 2017A Survey on Content-Aware Video Analysis for Sportscited by 27
- 2019Relational Action Forecastingcited by 27
- 2018Sports Camera Calibration via Synthetic Datacited by 26
- 2018Pointly-Supervised Action Localizationcited by 12
- 2018Learning Sports Camera Selection from Internet Videoscited by 6
- 2014Automatic Summarization of Soccer Highlights Using Audio-visual Descriptorscited by 5
- 2019Deep Metric Learning with Density Adaptivitycited by 3
- 2020Robust Visual Object Tracking with Two-Stream Residual Convolutional Networkscited by 3
Built on by
Citing papers per year: 2022: 2; 2023: 1; 2024: 3; 2025: 1.
7 papers.
- 2022Wave-ViT: Unifying Wavelet and Transformers for Visual Representation Learningcited by 78
- 2023Ball Trajectory Inference from Multi-Agent Sports Contexts Using Set Transformer and Hierarchical Bi-LSTMcited by 12
- 2024A Reinforcement Learning-Based Automatic Video Editing Method Using Pre-trained Vision-Language Modelcited by 8
- 2022LTC-SUM: Lightweight Client-driven Personalized Video Summarization Framework Using 2D CNNcited by 4
- 2024Pseudo Dataset Generation for Out-of-Domain Multi-Camera View Recommendationcited by 2
- 2024A multi-purpose automatic editing system based on lecture semantics for remote educationcited by 0
- 2025EditIQ: Automated Cinematic Editing of Static Wide-Angle Videos via Dialogue Interpretation and Saliency Cuescited by 0
- 2025EditIQ: Automated Cinematic Editing of Static Wide-Angle Videos via Dialogue Interpretation and Saliency Cues4 Feb 2025
- 2024A Reinforcement Learning-Based Automatic Video Editing Method Using Pre-trained Vision-Language Model7 Nov 2024
- 2024A multi-purpose automatic editing system based on lecture semantics for remote education7 Nov 2024
- 2024Pseudo Dataset Generation for Out-of-Domain Multi-Camera View Recommendation17 Oct 2024
- 2023Ball Trajectory Inference from Multi-Agent Sports Contexts Using Set Transformer and Hierarchical Bi-LSTM14 Jun 2023
- 2022Wave-ViT: Unifying Wavelet and Transformers for Visual Representation Learning11 Jul 2022
- 2022LTC-SUM: Lightweight Client-driven Personalized Video Summarization Framework Using 2D CNN22 Jan 2022