Insights from ECCV 2020 China Pre‑Conference Roundtable on the Academia‑Industry Gap in Computer Vision
At the ECCV 2020 China pre‑conference round‑table, leading academic and industry experts examined the narrowing yet still significant gap between research and deployment in computer vision, emphasizing philosophical thinking, T‑shaped expertise, product‑oriented perspectives, and collaborative skills, while highlighting growing joint publications and future priorities such as explainability, efficient learning, hardware‑software co‑design, and applications in surveillance, autonomous driving, and new‑retail.
On July 31, 2020, the Chinese Society of Image and Graphics (CSIG) organized the ECCV 2020 China Pre‑Conference, featuring a round‑table discussion titled “Computer Vision: How Wide Is the Gap Between Academia and Industry?”. The panel included leading experts from academia and industry such as Dr. Dong Jing (CAS), Dr. Hua Gang (Wormpex AI), Dr. Tian Qi (Huawei Cloud), Dr. Wang Jingdong (Microsoft Research Asia), Dr. Wei Xiaolin (Meituan AI), and Prof. Xiong Hongkai (Shanghai Jiao‑Tong University).
The discussion highlighted several recurring themes:
Philosophical thinking: Academic researchers should cultivate a philosophical, reflective mindset to pursue deeper truths beyond mere technical implementation.
“T‑shaped” knowledge: Graduates need deep expertise in a specific area combined with broad interdisciplinary understanding.
Product‑oriented perspective: Successful talent can view problems from a customer’s angle, identifying real‑world needs and translating them into research directions.
Collaboration and communication: Writing, proposal preparation, and effective teamwork are essential for both academic and industrial research.
Panelists described the “usage gap” as the transition from 0→1 (basic research) to 1→n (industrial deployment). They agreed that the gap is shrinking due to large‑scale data, compute resources, and talent flow between sectors, yet emphasized that both a tighter gap and occasional larger gaps can drive innovation.
Regarding conference submissions, the panel noted the growing number of joint academia‑industry papers at ECCV 2020 (104 oral papers out of 1400 submissions) and stressed the mutual benefits of shared datasets and joint projects.
Future research trends identified include:
Improving explainability, safety, robustness, and transparency of deep‑learning‑based vision systems.
Efficient learning with small data, self‑supervised methods, knowledge distillation, and model compression.
Hardware‑software co‑design, edge computing, and AutoML to reduce development cost.
Application hotspots: security surveillance, autonomous driving, and offline retail (new‑retail).
The panel concluded that academia and industry complement each other like two sides of a coin; continuous dialogue and talent exchange are crucial for a harmonious and innovative future in computer vision.
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
Meituan Technology Team
Over 10,000 engineers powering China’s leading lifestyle services e‑commerce platform. Supporting hundreds of millions of consumers, millions of merchants across 2,000+ industries. This is the public channel for the tech teams behind Meituan, Dianping, Meituan Waimai, Meituan Select, and related services.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.