Automated animal tracking software has revolutionized behavioral studies, particularly in tracking laboratory creatures such as aquarium fish, which is fundamental in neuroscience, medicine, and biomechanics. Despite advances, current open source tracking tools often need more precision under various conditions, especially when faced with obstacles or complex environments.
Commonly used tracking solutions employ techniques such as background subtraction or speckle detection, facing limitations in natural environments or aquariums due to reflections, waves, and dynamic backgrounds. While specialized software for specific fish models, such as zebrafish, works well under typical conditions, it struggles in a variety of practical scenarios due to limitations inherent to the method.
To address these challenges, a UK-based research team introduced a hybrid method, fusing deep learning and traditional computer vision techniques to improve the accuracy of fish tracking in complex experiments.
Unlike background subtraction or speckle detection techniques, the proposed new technique employs adaptive object detection using deep learning, enabling precise tracking of the Picasso triggerfish amidst different backgrounds, occlusions or deformations. By integrating optical flow computation with object detection and tracking, this approach ensures robustness to changes in fish appearance or occlusion by obstacles, providing accurate trajectory information despite the complex conditions that often occur. they challenge basic methods such as background subtraction or speckle detection.
This innovative approach combines the adaptability of deep learning with the precision of classical vision in centroid tracking, providing a more robust solution for monitoring fish behavior in challenging environments.
Their paper describes a pioneering method for analyzing Picasso triggerfish behavior using video processing in controlled tank environments. He uses a GoPro Hero 5 camera and advanced tools like EfficientDet and optical flow techniques.
The deep learning part involves the use of object detection and tracking. Specifically, the paper uses a deep learning-based object detector (EfcientDet) to identify both the Picasso triggerfish and cylindrical obstacles in the video frames. This detector is retrained to accurately detect these specific objects within the video data.
On the other hand, traditional computer vision techniques are used in the tracking process. The authors employ classical optical flow estimation between consecutive frames, a conventional method in computer vision, to estimate the trajectory and movement of fish and accurately identify fish trajectories amidst obstacles. By using optical flow between consecutive frames, they determine the movement of the fish, which helps understand how obstacles affect the behavior of the fish.
Initially, manual annotations on the videos trained the deep learning object detector, complemented by an object tracker to fill detection gaps using high-confidence nearby identifications.
Crucially, the method identified gates and areas between obstacles where fish move through, using Voronoi cell methods. For the doors that limit the tanks, imaginary obstacles were introduced to aid the identification of the doors.
Despite challenges such as partial occlusion and proximity to obstacles that affect accuracy, the method achieved a remarkable 97% alignment between calculated and manual fish trajectories. The researchers published their software, dataset, and tutorial under a Creative Commons license, supporting the broader scientific community in using computer vision tools for animal tracking.
However, adapting this method to complex scenarios or multiple animals might require further refinement, considering challenges such as partial occlusion or intricate environments.
In summary, this innovative fusion of deep learning and traditional computer vision techniques significantly improves the accuracy of tracking animals, particularly fish in complex experimental setups. While impressive results are achieved, challenges remain that require further refinement for broader applications beyond controlled environments. The published assets and tutorial provide crucial resources for potential adaptations and advancements in automated animal tracking.
Review the Prepress Paper. All credit for this research goes to the researchers of this project. Also, don't forget to join. our SubReddit of more than 35,000 ml, 41k+ Facebook community, Discord channel, LinkedIn Grabove, Twitterand Electronic newsletterwhere we share the latest news on ai research, interesting ai projects and more.
If you like our work, you'll love our newsletter.
Mahmoud is a PhD researcher in machine learning. He also owns a
Bachelor's degree in Physical Sciences and Master's degree in
telecommunications systems and networks. Your current areas of
The research concerns computer vision, stock market prediction and depth.
learning. He produced several scientific articles on the relationship of people.
identification and study of the robustness and stability of depths
networks.
<!– ai CONTENT END 2 –>
