The 1st edition of MAI: Mobile AI workshop will be held on June, 2021 in conjunction with CVPR 2021.
Camera scene detection task is one of the most popular problems related to mobile photography. While nowadays almost every smartphone has a built-in automatic mode for this task, there are still no free open-source datasets and models that can solve this problem accurately and fast. To address this problem, we propose a novel dataset consisting of 30 most important camera scene classes and organize a camera scene detection challenge. In this competition, the target is to obtain high precision results as measured by Top-1 / Top-3 accuracy, and at the same time the model should be very efficient as measured by inference time on the mobile platform.
The top ranked participants will be awarded and invited to follow the CVPR submission guide for workshops to describe their solution and to submit to the associated MAI workshop at CVPR 2021.
More details are found on the data section of the competition.
To learn more about the competition, to participate in the challenge, and to access the dataset with low and high resolution image pairs everybody is invited to register.
The training data is already made available to the registered participants.
We have opened a new AI Benchmark forum, where the participants can ask and discuss (highly recommended!) any questions related to MAI 2021 workshop and challenges.
For any private questions, you can contact the organizers directly by email (Andrey Ignatov (andrey [at] vision.ee.ethz.ch), Grigory Malivenko (nerox8664 [at] gmail.com), and Radu Timofte (Radu.Timofte [at] vision.ee.ethz.ch)). Please note that all posts left in the Codalab forum will not be checked.
In this challenge, each submission is validated based on the following two metrics:
1. The accuracy of the predictions.
2. The runtime of the model on the actual target mobile platform.
The exact scoring formula used in this challenge is provided below:
where C is a constant normalization factor that does not depend on the submission. Please note that we might additionally award submissions showing extraordinary fidelity or runtime results even if they are not top ranked according to the above formula.
The evaluation of the model prediction consists from computing top-1 and top-3 classification accuracy on the test images. The description of the runtime evaluation can be found in the corresponding tab. We report the average results over all the processed images belonging to the evaluation dataset.
For submitting the results, you need to follow these steps:
Runtime per image [ms] : 10.43
Extra Data [1] / No Extra Data [0] : 1
Mobile Device : Samsung Galaxy S10 (Exynos)
Acceleration : NNAPI
Other description : Solution is based on the MobileNet-V2 architecture. The model was pre-trained on the ImageNet dataset.
http://data.vision.ee.ethz.ch/ihnatova/Factsheet_Template_MAI2021_Challenges.zip
These are the official rules (terms and conditions) that guvern how the Mobile 2021 challenge on Real Image Denoising will operate. This challenge will be simply reffered to as the "challenge" or the "contest" throghout the remaining part of these rules and may be named as "MAI" or "Denoising" benchmark, challenge, or contest, elsewhere (our webpage, our documentation, other publications).
In these rules, "we", "our", and "us" refer to the organizers (Andrey Ignatov (andrey [at] vision.ee.ethz.ch), Grigory Malivenko (nerox8664 [at] gmail.com) and Radu Timofte (Radu.Timofte [at] vision.ee.ethz.ch)) of MAI challenge and "you" and "yourself" refer to an eligible contest participant.
Note that these official rules can change during the contest until the start of the final phase. If at any point during the contest the registered participant considers that can not anymore meet the eligibility criteria or does not agree with the changes in the official terms and conditions then it is the responsability of the participant to send an email to the organizers such that to be removed from all the records. Once the contest is over no change is possible in the status of the registered participants and their entries.
This is a skill-based contest and chance plays no part in the determination of the winner (s).
The goal of the contest is to learn a mapper from an input noisy image to an output clean image and the challenge is called Real Image Denoising.
Focus of the contest: it will be made available a new dataset adapted for the specific needs of the challenge. The images have a large diversity of contents. We will refer to this dataset, its partition, and related materials as Camera Scene Classification Dataset. The dataset is divided into training, validation and testing data. We focus on quality of the results, the aim is to achieve high Top-1 and Top-3 accuracy on the validation / test images. The participants will not have access to the ground truth labels for validation and test data. The ranking of the participants is according to the performance of their methods on the test data and on the mobile platform. The participants will provide descriptions of their methods, details on (run)time complexity, platform and (extra) data used for modeling. The winners will be determined according to their entries, the reproducibility of the results and uploaded codes or executables, and the above mentioned criteria as judged by the organizers.
The registered participants will be notified by email if any changes are made to the schedule. The schedule is available on the Mobile AI workshop web page and on the Overview of the Codalab competition.
You are eligible to register and compete in this contest only if you meet all the following requirements:
This contest is void wherever it is prohibited by law.
Entries submitted but not qualified to enter the contest, it is considered voluntary and for any entry you submit MAI reserves the right to evaluate it for scientific purposes, however under no circumstances will such entries qualify for sponsored prizes. If you are an employee, affiliated with or representant of any of the MAI challenge sponsors then you are allowed to enter in the contest and get ranked, however, if you will rank among the winners with eligible entries you will receive only a diploma award and none of the sponsored money, products or travel grants.
NOTE: industry and research labs are allowed to submit entries and to compete in both validation phase and final test phase. However, in order to get officially ranked on the final test leaderboard and to be eligible for awards the reproducibility of the results is a must and, therefore, the participants need to make available and submit their codes or executables. All the top entries will be checked for reproducibility and marked accordingly.
We will have 3 categories of entries in the final test ranking:
1) checked with publicly released codes
2) checked with publicly released executable
3) unchecked (with or without released codes or executables)
In order to be eligible for judging, an entry must meet all the following requirements:
Entry contents: the participants are required to submit image results and code or executables. To be eligible for prizes, the top ranking participants should publicly release their code or executables under a license of their choice, taken among popular OSI-approved licenses (http://opensource.org/licenses) and make their code or executables online accessible for a period of not less than one year following the end of the challenge (applies only for top three ranked participants of the competition). To enter the final ranking the participants will need to fill out a survey (fact sheet) briefly describing their method. All the participants are also invited (not mandatory) to submit a paper for peer-reviewing and publication at the MAI Workshop and Challenges (to be held online on June, 2021). To be eligible for prizes, the participants score must improve the baseline performance provided by the challenge organizers.
Use of data provided: all data provided by MAI are freely available to the participants from the website of the challenge under license terms provided with the data. The data are available only for open research and educational purposes, within the scope of the challenge. MAI and the organizers make no warranties regarding the database, including but not limited to warranties of non-infringement or fitness for a particular purpose. The copyright of the images remains in property of their respective owners. By downloading and making use of the data, you accept full responsibility for using the data. You shall defend and indemnify MAI and the organizers, including their employees, Trustees, officers and agents, against any and all claims arising from your use of the data. You agree not to redistribute the data without this notice.
Other than what is set forth below, we are not claiming any ownership rights to your entry. However, by submitting your entry, you:
Are granting us an irrevocable, worldwide right and license, in exchange for your opportunity to participate in the contest and potential prize awards, for the duration of the protection of the copyrights to:
Agree to sign any necessary documentation that may be required for us and our designees to make use of the rights you granted above;
Understand and acknowledge that us and other entrants may have developed or commissioned materials similar or identical to your submission and you waive any claims you may have resulting from any similarities to your entry;
Understand that we cannot control the incoming information you will disclose to our representatives or our co-sponsor’s representatives in the course of entering, or what our representatives will remember about your entry. You also understand that we will not restrict work assignments of representatives or our co-sponsor’s representatives who have had access to your entry. By entering this contest, you agree that use of information in our representatives’ or our co-sponsor’s representatives unaided memories in the development or deployment of our products or services does not create liability for us under this agreement or copyright or trade secret law;
Understand that you will not receive any compensation or credit for use of your entry, other than what is described in these official rules.
If you do not want to grant us these rights to your entry, please do not enter this contest.
The participants will follow the instructions on the CodaLab website to submit entries
The participants will be registered as mutually exclusive teams. Each team is allowed to submit only one single final entry. We are not responsible for entries that we do not receive for any reason, or for entries that we receive but do not work properly.
The participants must follow the instructions and the rules. We will automatically disqualify incomplete or invalid entries.
The board of MAI will select a panel of judges to judge the entries; all judges will be forbidden to enter the contest and will be experts in causality, statistics, machine learning, computer vision, or a related field, or experts in challenge organization. A list of the judges will be made available upon request. The judges will review all eligible entries received and select (three) winners for each or for both of the competition tracks based upon the prediction score on test data. The judges will verify that the winners complied with the rules, including that they documented their method by filling out a fact sheet.
The decisions of these judges are final and binding. The distribution of prizes according to the decisions made by the judges will be made within three (3) months after completion of the last round of the contest. If we do not receive a sufficient number of entries meeting the entry requirements, we may, at our discretion based on the above criteria, not award any or all of the contest prizes below. In the event of a tie between any eligible entries, the tie will be broken by giving preference to the earliest submission, using the time stamp of the submission platform.
The financial sponsors of this contest are listed on Mobile AI 2021 workshop web page . There will be economic incentive prizes and travel grants for the winners (based on availability) to boost contest participation; these prizes will not require participants to enter into an IP agreement with any of the sponsors, to disclose algorithms, or to deliver source code to them. The participants affiliated with the industry sponsors agree to not receive any sponsored money, product or travel grant in the case they will be among the winners.
Incentive Prizes for each track competitions (tentative, the prizes depend on attracted funds from the sponsors)
Publishing papers is optional and will not be a condition to entering the challenge or winning prizes. The top ranking participants are invited to submit a paper following CVPR2021 author rules, for peer-reviewing to MAI workshop.
The results of the challenge will be published together with MAI 2021 workshop papers in the 2021 CVPR Workshops proceedings.
The top ranked participants and participants contributing interesting and novel methods to the challenge will be invited to be co-authors of the challenge report paper which will be published in the 2021 CVPR Workshops proceedings. A detailed description of the ranked solution as well as the reproducibility of the results are a must to be an eligible co-author.
If there is any change to data, schedule, instructions of participation, or these rules, the registered participants will be notified on the competition page and/or at the email they provided with the registration.
Within seven days following the determination of winners we will send a notification to the potential winners. If the notification that we send is returned as undeliverable, or you are otherwise unreachable for any reason, we may award the prize to an alternate winner, unless forbidden by applicable law.
The prize such as money, product, or travel grant will be delivered to the registered team leader given that the team is not affiliated with any of the sponsors. It is up to the team to share the prize. If this person becomes unavailable for any reason, the prize will be delivered to be the authorized account holder of the e-mail address used to make the winning entry.
If you are a potential winner, we may require you to sign a declaration of eligibility, use, indemnity and liability/publicity release and applicable tax forms. If you are a potential winner and are a minor in your place of residence, and we require that your parent or legal guardian will be designated as the winner, and we may require that they sign a declaration of eligibility, use, indemnity and liability/publicity release on your behalf. If you, (or your parent/legal guardian if applicable), do not sign and return these required forms within the time period listed on the winner notification message, we may disqualify you (or the designated parent/legal guardian) and select an alternate selected winner.
Following the previous NTIRE and AIM challenges, we accept multiple submissions from a single team if they are significantly different. However, solutions with minor differences can be rejected by the organizers. For example, if solution A and B are submitted where B is an ensemble method of A, we do not consider them to be essentially different.
If you are preparing multiple submissions, please contact the organizers in advance for clarification.
The terms and conditions are inspired by and use verbatim text from the `Terms and conditions' of ChaLearn Looking at People Challenges and of the NTIRE 2017, 2018, 2019, 2020 and 2021 challenges and of the AIM 2019 and 2020 challenges .
In this challenge, each submission is validated based on the following two metrics:
1. The accuracy of the predictions.
2. The runtime of the model on the actual target mobile platform.
To run your model on a mobile device, it has first to be converted to TensorFlow Lite (TFLite) format. Please find the conversion instructions for TensorFlow and Keras models below.
If you haven't worked with TensorFlow Lite and network quantization before, try to get a simple floating-point TFLite model first. This is fairly easy, you can just use the following scripts:
Kerass & TensorFlow: https://github.com/aiff22/MAI-2021-Workshop/blob/main/classification_quantized/keras_to_tflite_float.py
You can then try to run the obtained model on your own smartphone using the instructions provided in the next section. Please note that there is no straightforward way to get a quantized TFLite network from a PyTorch model, therefore we strongly recommend you to use either TensorFlow or Keras for all development.
The easiest way to get a quantized TFLite model is to apply TensorFlow's post-training quantization to your pre-trained floating-point network. In this case, you don't need to do any model retraining, all quantization and conversion instructions are provided below:
Keras & TensorFlow: https://github.com/aiff22/MAI-2021-Workshop/blob/main/classification_quantized/keras_to_tflite_quant.py
Please note that you should generally use the entire training dataset for getting quantized tensor stats (check the representative_dataset() function).
Don't forget to always test the outputs of the resulting TFLite quantized network on several input images, for this you can use the following script:
https://github.com/aiff22/MAI-2021-Workshop/blob/main/classification_quantized/test_tflite_model.py
You can potentially get a more accurate quantized model by fine-tuning your pre-trained floating-point network with quantization aware training. This can be done quite easily, you can find some useful examples below:
https://www.tensorflow.org/model_optimization/guide/quantization/training_example
After you obtained a TFLite model, you can check its runtime using the following two options:
a. Run it remotely on the target evaluation platform (the instructions will be announced soon).
b. Run it on your own smartphone.
We strongly encourage you to do all main testing on your own device since:
1) If it is not working on your smartphone, then it will also not work on the target platform.
2) It is much more convenient and efficient to debug / profile the model locally.
3) You have an unlimited number of runs on your own device.
To run the resulting TFLite model on your smartphone, you can use AI Benchmark application allowing you to execute the model on mobile 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.
That's it - you should see the runtime, initialization time and RAM consumption of your model. Now you can try to play with different TF ops, layers and resolutions to improve all these parameters and get a very fast and efficient network!
You can upload your TFLite model and check its runtime on an iPhone with the Apple Bionic SoC instantaneously using the following link: http://lightspeed.difficu.lt:60002/
Please note that:
1. You need to enter the same email address that you used for registering in the Codalab competition.
2. To avoid large queues, there is a 3 minute delay between two subsequent submissions.
Each time you try a new model architecture - try to initialize it with random weights, then convert it to TFLite and run on your own device using AI Benchmark. While this will take you only a few minutes, you will be sure that the model you are developing can be successfully converted and executed on a mobile device, and besides that you will get its estimated runtime compared to your previous entry.
The runtime of your final model will be evaluated on the Apple Bionic platform. In general, there are no specific requirements here, therefore you should try to optimize the runtime of the model on your own device when running it with NNAPI.
The Mobile AI challenge on real image denoising is organized jointly with the Mobile AI 2021 workshop. The results of the challenge will be published at Mobile AI 2021 workshop and in the CVPR 2021 Workshops proceedings.
Andrey Ignatov (andrey [at] vision.ee.ethz.ch), Grigory Malivenko (nerox8664 [at] gmail.com) and Radu Timofte (Radu.Timofte [at] vision.ee.ethz.ch) are the contact persons and direct managers of the Mobile AI challenge.
More information about Mobile AI workshop and challenge organizers is available here: http://ai-benchmark.com/workshops/mai/2021/
Start: Jan. 5, 2021, midnight
Description: Development phase - submit the results on the validation data.
Start: March 15, 2021, 11:59 p.m.
Description: Testing phase - submit the results on the test data.
March 21, 2021, 11:59 p.m.
You must be logged in to participate in competitions.
Sign In# | Username | Score |
---|---|---|
1 | sayannath | 97.6667 |
2 | pyt | 97.5000 |
3 | peilin | 97.5000 |