Over the past years, mobile AI-based applications are becoming more and more ubiquitous. Various deep learning models can now be found on any mobile device, starting from smartphones running portrait segmentation, image enhancement, face recognition and natural language processing models, to smart-TV boards coming with sophisticated image super-resolution algorithms. The performance of mobile NPUs and DSPs is also increasing dramatically, making it possible to run complex deep learning models and to achieve fast runtime in the majority of tasks.

While many research works targeted at efficient deep learning models have been proposed recently, the evaluation of the obtained solutions is usually happening on desktop CPUs and GPUs, making it nearly impossible to estimate the actual inference time and memory consumption on real mobile hardware. To address this problem, we introduce the first Mobile AI Workshop, where all deep learning solutions are developed for and evaluated on mobile devices.

Due to the performance of the last-generation mobile AI hardware, the topics considered in this workshop will go beyond the simple classification tasks, and will include such challenging problems as image denoising, HDR photography, accurate depth estimation, learned image ISP pipeline, real-time image and video super-resolution. All information about the challenges, papers, invited talks and workshop industry partners is provided below.

LIVE

Join the main workshop Zoom conference for a Q&A session: https://ethz.zoom.us/j/61668005465

SCHEDULE

Deploying Deep Learning Models on Mobile NPUs and Beyond:  What's New in 2025?


07:00 Pacific Time   ┈   Andrey Ignatov   ┈   AI Benchmark Project Lead, ETH Zurich


Abstract: In this tutorial, we will first recall all basic concepts, steps and optimizations required for efficient AI inference on mobile NPUs. Next, we will go into more detail about the latest mobile platforms from Qualcomm, MediaTek, Google, Samsung, Unisoc and Apple released during the past year, and will compare their inference speed when running common computer vision models. We will talk about power efficiency of mobile NPUs, and will analyze their energy consumption during typical AI workloads. Finally, we will go beyond Android and iOS, covering the topics of AI model deployment on NPUs in Windows, Linux and MacOS systems, analyzing available ML frameworks and their performance.

Recent Mobile AI Advances and Case Studies


09:00 Pacific Time   ┈   CM Cheng, YuSyuan Xu and Haoyun Chen   ┈   MediaTek Inc.


Abstract: In this talk, MediaTek will provide you with an overview of their software and hardware platforms and several mobile-oriented research topics. The first part of the talk would be devoted to the discussion of the recent mobile AI advances and MediaTek AI ecosystem. The second part will focus on two recent case studies: Vision Mamba and MOE, which are related to efficient model deployment of mobile AI hardware.

Quantized Image Super-Resolution on Mobile NPUs:  Results and Top Solutions

09:40 Pacific Time   ┈   Andrey Ignatov, ETH Zurich

Evaluation Platform:
Google Tensor NPU

09:50 Pacific Time     RepNet-VSR: Reparameterizable Architecture for High-Fidelity Video Super-Resolution


Biao Wu, Diankai Zhang, Shaoli Liu, Si Gao, Chengjian Zheng, Ning Wang


☉  ZTE Corporation


10:05 Pacific Time     CDVS: Compressed Domain On Device Memory Efficient 8K Video SlowMo


Jing Li, Chengyu Wang, Hamid Sheikh, SeokJun Lee


☉  Samsung Research America & Purdue University


Learned Smartphone ISP on Mobile GPUs Challenge:  Results and Top Solutions

10:20 Pacific Time   ┈   Andrey Ignatov, ETH Zurich

Evaluation Platform:
Snapdragon 8 Elite (Adreno GPU) / Dimensity 9400 (Mali GPU)

10:30 Pacific Time     Learned Lightweight Smartphone ISP with Unpaired Data


Andrei Arhire, Radu Timofte


☉  Alexandru Ioan Cuza University


10:45 Pacific Time     Compressed Domain Multiframe Processing


Chengyu Wang, Jing Li, Saurabh Kumar, Seok-Jun Lee, Hamid Sheikh


☉  Samsung Research America & SRI-Bangalore



11:00 Pacific Time     Break & Lunch



RGB Photo Enhancement on Mobile GPUs Challenge:  Results and Top Solutions

12:00 Pacific Time   ┈   Andrey Ignatov, ETH Zurich

Evaluation Platform:
Snapdragon 8 Elite (Adreno GPU) / Dimensity 9400 (Mali GPU)

12:15 Pacific Time     PETAH: Parameter Efficient Task Adaptation for Hybrid Transformers


Syed Shakib Sarwar, Mostafa Elhoushi, Maximilian Augusting, Yuecheng Li, Sai Zhang, Barbara De Salvo


☉  Meta Inc. & IEEE & University of Tubingen & New York University


12:30 Pacific Time     ActNAS: Generating Efficient YOLO Models using Activation NAS


Sudhakar Sah, Ravish Kumar, Darshan Ganji, Ehsan Saboori


☉  Deeplite


12:45 Pacific Time     FLAR-SVD: Fast and Latency-Aware Singular Value Decomposition for Model Compression


Moritz Thoma, Jorge Villasante, Emad Aghajanzadeh, Shambhavi Sampath, Pierpaolo Mori et al.


☉  BMW Group & Technical University of Munich


13:15 Pacific Time     Robust 6DoF Pose Estimation Against Depth Noise and Evaluation on a Mobile Dataset


Zixun Huang, Keling Yao, Zhihao Zhao, Chuanyu Pan, Allen Yang


☉  UC Berkeley & Carnegie Mellon University & University of California


13:30 Pacific Time     Cycle Training with Semi-Supervised Domain Adaptation for Real-Time Mobile Scene Detection


Huu-Phong Phan-Nguyen, Anh Dao, Tien-Huy Nguyen, Tuan Quang et al.


☉  University of Infomation Technology & Michigan State University & LPL Financial Corp et al.


13:45 Pacific Time     RepFC: Universal Structural Reparametrization Block for High Performance


Shambhavi Balamuthu Sampath, Judeson Anthony Fernando, Moritz Thoma, Nael Fasfous et al.


☉  BMW Group & Technical University of Munich





14:00 Pacific Time     Discussion & Wrap Up


CALL FOR PAPERS

Being a part of CVPR 2025, we invite the authors to submit high-quality original papers proposing various machine learning based solutions for mobile, embedded and IoT platforms. The topics of interest cover all major aspects of AI and deep learning research for mobile devices including, but not limited to:

•   Efficient deep learning models for mobile devices

•   Image / video super-resolution on low-power hardware

•   General smartphone photo and video enhancement

•   Deep learning applications for mobile camera ISPs

•   Fast image classification / object detection algorithms

•   Real-time semantic image segmentation

•   Image or sensor based identity recognition

•   Activity recognition using smartphone sensors

•   Depth estimation w/o multiple cameras

•   Portrait segmentation / bokeh effect rendering

•   Perceptual image manipulation on mobile devices

•   NLP models optimized for mobile inference

•   Artifacts removal from mobile photos / videos

•   RAW image and video processing

•   Low-power machine learning inference

•   Machine and deep learning frameworks for mobile devices

•   AI performance evaluation of mobile and IoT hardware

•   Industry-driven applications related to the above problems

To ensure high quality of the accepted papers, all submissions will be evaluated by research and industry experts from the corresponding fields. All accepted workshop papers will be published in the CVPR 2025 Workshop Proceedings by Computer Vision Foundation Open Access and IEEE Xplore Digital Library. The authors of the best selected papers will be invited to present their work during the actual workshop event at CVPR 2025.

The detailed submission instructions and guidelines can be found here.

SUBMISSION DETAILS @ CVPR





For AIM challenge papers submission @ ICCV, please refer to these instructions.

Format and paper length A paper submission has to be in English, in pdf format, and at most 8 pages (excluding references) in double column. The paper format must follow the same guidelines as for all CVPR 2025 submissions: https://cvpr.thecvf.com/Conferences/2025/AuthorGuidelines
Author kit The author kit provides a LaTeX2e template for paper submissions. Please refer to this kit for detailed formatting instructions: https://github.com/cvpr-org/author-kit/archive/refs/tags/CVPR2025-v3.1(latex).zip
Double-blind review policy The review process is double blind. Authors do not know the names of the chair / reviewers of their papers. Reviewers do not know the names of the authors.
Dual submission policy Dual submission is allowed with CVPR2025 main conference only. If a paper is submitted also to CVPR and accepted, the paper cannot be published both at the CVPR and the workshop.
Proceedings Accepted and presented papers will be published after the conference in CVPR Workshops proceedings together with the CVPR 2025 main conference papers.
Submission site https://cmt3.research.microsoft.com/MAI2025

TIMELINE @ ICCV

Workshop Event Date   [ 5pm Pacific Time, 2025 ]
Website onlineJanuary 25
Paper submission server online February 13
Paper submission deadline [challenge papers] July 9
Paper decision notification July 11
Camera ready deadline August
Workshop day October (TBA)

TIMELINE @ CVPR

Workshop Event Date   [ 5pm Pacific Time, 2025 ]
Website onlineJanuary 25
Paper submission server online February 13
Paper submission deadline [early submission] March 10
Paper decision notification [early submission] March 31
Paper submission deadline [late submission & challenge papers] March 28
Paper decision notification [late submission] March 31
Camera ready deadline April 7
Workshop day June (TBA)

DEEP LEARNING ON MOBILE DEVICES: TUTORIAL

Have some questions?  Leave them on the AI Benchmark Forum

RUNTIME VALIDATION

In each MAI 2025 challenge track, the participants have a possibility to check the runtime of their solutions remotely on the target platforms. For this, the converted TensorFlow Lite models should be uploaded to a special web-server, and their runtime on the actual target devices will be returned instantaneously or withing 24 hours, depending on the track. The detailed model conversion instructions and links can be found in the corresponding challenges.

Besides that, we strongly encourage the participants to check the speed and RAM consumption of the obtained models locally on your own Android devices. This will allow you to perform model profiling and debugging faster and much more efficiently. To do this, one can use AI Benchmark application allowing you to load a custom TFLite model and run it with various acceleration options, including CPU, GPU, DSP and NPU:

1.  Download AI Benchmark from the Google Play / website and run its standard tests.
2.  After the end of the tests, enter the PRO Mode and select the Custom Model tab there.
3.  Rename the exported TFLite model to model.tflite and put it into the Download folder of your device.
4.  Select your mode type, the desired acceleration / inference options and run the model.

You can find the screenshots demonstrating these 4 steps below:

CONTACTS

Andrey Ignatov

Computer Vision Lab

ETH Zurich, Switzerland

andrey@vision.ee.ethz.ch

Radu Timofte

Computer Vision Laboratory

University of Würzburg, Germany

radu.timofte@uni-wuerzburg.de

Computer Vision Laboratory, ETH Zurich

Switzerland, 2025