ICCV 2025 Announcements! 24% Acceptance Rate — Did You Secure Your Ticket to Hawaii?}

ICCV 2025 will take place from October 19 to 25 in Hawaii, USA. Recently, the ICCV official sent notifications to authors regarding the acceptance results for this year’s papers.

Data shows that this year, the conference received a total of 11,239 valid submissions, all of which entered the review process. The program committee recommended 2,699 papers for acceptance, resulting in an acceptance rate of 24%.

Compared to previous years, the submission volume in 2025 is nearly three times that of 2019, reflecting the rapid expansion of the computer vision field and the increasing vibrancy of academic research.

• ICCV 2023: 8,260 submissions, 2,160 accepted, acceptance rate approximately 26.15%.
• ICCV 2021: 6,152 submissions, 1,612 accepted, acceptance rate 26.20%.
• ICCV 2019: 4,323 submissions, 1,075 accepted, acceptance rate 25%.

Despite the significant increase in submissions, ICCV’s acceptance rate has remained relatively stable over the past few years, generally staying within 25% to 26%.

Following CVPR 2025, ICCV 2025 also implemented a new policy aimed at strengthening accountability and integrity. The program committee identified 25 reviewers with irresponsible conduct, leading to the rejection of 29 papers associated with them.

Among these rejected papers, 12 would have been accepted if not for the misconduct, sparking some controversy.

Known accepted papers include, but are not limited to:

• High-fidelity 3D Geometry Generation from Images via Normal Bridging

• Mind the Cost of Scaffold! Benign Clients May Even Become Accomplices of Backdoor Attack

• UKBOB: One Billion MRI Labeled Masks for Generalizable 3D Medical Image Segmentation

• FlowR: Flowing from Sparse to Dense 3D Reconstructions

• Magic Insert: Style-Aware Drag-and-Drop

• OD-RASE: Ontology-Driven Risk Assessment and Safety Enhancement for Autonomous Driving

• UniVG: A Generalist Diffusion Model for Unified Image Generation and Editing

ICCV (International Conference on Computer Vision) is a biennial academic conference organized by IEEE, alongside CVPR and ECCV, forming the top three in the field of computer vision. The conference typically lasts four to five days, featuring tutorials, technical sessions, poster presentations, workshops, and exhibitions.

Challenges from the Surge in Submissions

Since AlexNet’s breakthrough in ImageNet in 2012, deep learning has revolutionized computer vision, NLP, and speech recognition. The rapid iteration of technology and emergence of new applications have provided abundant research topics. Recently, disruptive technologies like LLMs and generative AI have further fueled global research enthusiasm, leading to a surge in papers.

As submission volumes grow, peer review processes face unprecedented challenges. Some conferences have seen over 10,000 submissions, with NIPS (NeurIPS) potentially exceeding 30,000 submissions this year. Issues of review fairness and reviewer responsibility are increasingly prominent.

Previous controversies over review fairness are documented, with solutions proposed in research papers, such as: Review fairness issues and proposed solutions.

• Number of AI conference submissions continues to rise, with over 10,000 papers submitted to major conferences, and more than 30,000 expected at NeurIPS.

Rapid technological iteration and new application scenarios have generated numerous research topics. The emergence of LLMs and generative AI has further ignited global enthusiasm, resulting in a large volume of research papers.

With the surge in submissions, peer review systems are under immense pressure. Some conferences have faced disputes over review fairness, with calls for reforms to improve transparency and accountability.