During this Mosquito season, we officially welcomed AIMA, which is the name assigned to the artificial intelligence of Mosquito Alert (Artificial Intelligence Mosquito Alert).
AIMA has been developed with the aim of identifying mosquito images in real time and providing an almost immediate response to both community users and public health managers.
Like all artificial intelligence systems, AIMA learns through training, and in this case, the system has been trained by the community of artificial intelligence specialists at Mosquito Alert, using the photos that have been collected through the app over the years. Therefore, AIMA is specifically trained to detect specific mosquito species that are the subject of study in the Mosquito Alert project.
From now on, every time a mosquito photo is uploaded to the application, this information will be processed using a photo classification system. This classification will be later confirmed by the network of specialists in digital entomology.
The challenge of identifying small objects
Detecting small objects is one of the most challenging problems in computer vision. This is the challenge that AIMA faces when mosquito images are received through the Mosquito Alert App. Accurately detecting and localizing objects that are relatively small in size compared to the overall image poses a significant challenge.
However, identifying mosquito species contributes to early intervention and effective disease management. By leveraging machine learning and deep learning techniques, our goal is to automate the labor-intensive image validation process, making mosquito identification fast, more efficient and accurate. This helps the experts save time and allows them to focus on the more challenging images.
In collaboration with AI CROWD, we present the Mosquito Identification Challenge 2023—an opportunity to directly impact public health initiatives. The challenge entails the development of state-of-the-art computer vision algorithms that can locate mosquitoes with pinpoint accuracy, despite their diminutive size, within the given images. With this competition, we aim to reach out to the community of artificial intelligence experts to see if they can contribute a solution that enhances the accuracy of AIMA.
The Mosquito Identification Challenge
The competition centred around utilising advanced computer vision techniques to detect and classify small objects, mosquito images. Participants will develop cutting-edge AI solutions that precisely identify mosquitoes within images captured by citizen contributors using their mobile devices.
The challenge started the 20th of June, and now it’is in the scond pase. The winners will be announced the 10th November 2023. The challenge prize includes a Travel Grant of 2500 CHF per team and conference access to the Applied Machine Learning Days (AMLD), with travel and accommodation expenses covered. More information can be found, here.
Second Phase Kick-off: Online Event on August 24th, 2023
Just as Round 2 of the challenge kicks-off, next 24th august we invite all AICROWD challenge participants to join the exclusive Mosquito Alert Townhall event . This is a great opportunity to learn from domain specialists, acquire insights, dive into innovative strategies, and have your burning questions addressed by a panel of experts.
Meet the speakers
- Frederic Bartumeus: as the codirector of Mosquito Alert and ICREA Research Professor and Head of the Computational and Theoretical Ecology Lab, he will shed light on the driving force and objectives behind this challenge.
- Roger Eritja: an experienced entomologist with a 40-year track on mosquito control, ecology and systematic working presently at CEAB-CSIC. Roger will unravel the complexities faced in digital entomology.
- Monika Faik: Our AI Research Technician will provide an in-depth walkthrough of the challenge and the dataset.
- Joan Garriga: A renowned Data Analyst and Scientist, Joan will share insights on current solutions and offer a sneak peek into what the subsequent round holds.
For more information, please visit the AICROWD challenge page, available here.