scholarly journals Automated Detection, Classification and Counting of Fish in Fish Passages With Deep Learning

2022 ◽  
Vol 8 ◽  
Author(s):  
Vishnu Kandimalla ◽  
Matt Richard ◽  
Frank Smith ◽  
Jean Quirion ◽  
Luis Torgo ◽  
...  

The Ocean Aware project, led by Innovasea and funded through Canada's Ocean Supercluster, is developing a fish passage observation platform to monitor fish without the use of traditional tags. This will provide an alternative to standard tracking technology, such as acoustic telemetry fish tracking, which are often not appropriate for tracking at-risk fish species protected by legislation. Rather, the observation platform uses a combination of sensors including acoustic devices, visual and active sonar, and optical cameras. This will enable more in-depth scientific research and better support regulatory monitoring of at-risk fish species in fish passages or marine energy sites. Analysis of this data will require a robust and accurate method to automatically detect fish, count fish, and classify them by species in real-time using both sonar and optical cameras. To meet this need, we developed and tested an automated real-time deep learning framework combining state of the art convolutional neural networks and Kalman filters. First, we showed that an adaptation of the widely used YOLO machine learning model can accurately detect and classify eight species of fish from a public high resolution DIDSON imaging sonar dataset captured from the Ocqueoc River in Michigan, USA. Although there has been extensive research in the literature identifying particular fish such as eel vs. non-eel and seal vs. fish, to our knowledge this is the first successful application of deep learning for classifying multiple fish species with high resolution imaging sonar. Second, we integrated the Norfair object tracking framework to track and count fish using a public video dataset captured by optical cameras from the Wells Dam fish ladder on the Columbia River in Washington State, USA. Our results demonstrate that deep learning models can indeed be used to detect, classify species, and track fish using both high resolution imaging sonar and underwater video from a fish ladder. This work is a first step toward developing a fully implemented system which can accurately detect, classify and generate insights about fish in a wide variety of fish passage environments and conditions with data collected from multiple types of sensors.

2022 ◽  
Vol 149 ◽  
pp. 106819
Author(s):  
Huazheng Wu ◽  
Xiangfeng Meng ◽  
Xiulun Yang ◽  
Xianye Li ◽  
Yongkai Yin

Nanophotonics ◽  
2021 ◽  
Vol 0 (0) ◽  
Author(s):  
Zhicheng Xiao ◽  
Andrea Alù

Abstract Fano resonances feature an asymmetric lineshape with controllable linewidth, stemming from the interplay between bright and dark resonances. They provide efficient opportunities to shape the scattering lineshape, but they usually lack flexibility and tunability and are hindered by loss in passive systems. Here, we explore a hybrid parity-time (PT) and anti-parity-time (APT) symmetric system supporting unitary scattering features with highly tunable Fano resonances. The PT-APT-symmetric system can be envisioned in nanophotonic and microwave circuit implementations, allowing for real-time control of the scattering lineshape and its underlying singularities. Our study shows the opportunities enabled by non-Hermitian platforms to control scattering lineshapes for a plethora of photonic, electronic, and quantum systems, with potential for high-resolution imaging, switching, sensing, and multiplexing.


2021 ◽  
Author(s):  
Tadayoshi Aoyama ◽  
Sarau Takeno ◽  
Kazuki Hano ◽  
Masaki Takasu ◽  
Masaru Takeuchi ◽  
...  

2015 ◽  
Vol 53 (8) ◽  
pp. 2693-2696 ◽  
Author(s):  
Ramzi Ghodbane ◽  
Shady Asmar ◽  
Marlena Betzner ◽  
Marie Linet ◽  
Joseph Pierquin ◽  
...  

Culture remains the cornerstone of diagnosis for pulmonary tuberculosis, but the fastidiousness ofMycobacterium tuberculosismay delay culture-based diagnosis for weeks. We evaluated the performance of real-time high-resolution imaging for the rapid detection ofM. tuberculosiscolonies growing on a solid medium. A total of 50 clinical specimens, including 42 sputum specimens, 4 stool specimens, 2 bronchoalveolar lavage fluid specimens, and 2 bronchial aspirate fluid specimens were prospectively inoculated into (i) a commercially available Middlebrook broth and evaluated for mycobacterial growth indirectly detected by measuring oxygen consumption (standard protocol) and (ii) a home-made solid medium incubated in an incubator featuring real-time high-resolution imaging of colonies (real-time protocol). Isolates were identified by Ziehl-Neelsen staining and matrix-assisted laser desorption ionization–time of flight mass spectrometry. Use of the standard protocol yielded 14/50 (28%)M. tuberculosisisolates, which is not significantly different from the 13/50 (26%)M. tuberculosisisolates found using the real-time protocol (P= 1.00 by Fisher's exact test), and the contamination rate of 1/50 (2%) was not significantly different from the contamination rate of 2/50 (4%) using the real-time protocol (P= 1.00). The real-time imaging protocol showed a 4.4-fold reduction in time to detection, 82 ± 54 h versus 360 ± 142 h (P< 0.05). These preliminary data give the proof of concept that real-time high-resolution imaging ofM. tuberculosiscolonies is a new technology that shortens the time to growth detection and the laboratory diagnosis of pulmonary tuberculosis.


2020 ◽  
Vol 32 (21) ◽  
pp. 1397-1400
Author(s):  
Beichen Fan ◽  
Fangzheng Zhang ◽  
Cong Ma ◽  
Yue Yang ◽  
Shilong Pan ◽  
...  

2019 ◽  
Vol 9 (1) ◽  
Author(s):  
Mirela Balan ◽  
Marta Trusohamn ◽  
Frank Chenfei Ning ◽  
Stefan Jacob ◽  
Kristian Pietras ◽  
...  

Abstract Preclinical trials of cancer drugs in animal models are important for drug development. The Rip1Tag2 (RT2) transgenic mouse, a model of pancreatic neuroendocrine tumours (PNET), has provided immense knowledge about PNET biology, although tumour progression occurs in a location inaccessible for real-time monitoring. To overcome this hurdle we have developed a novel platform for intravital 3D imaging of RT2 tumours to facilitate real-time studies of cancer progression. Pre-oncogenic islets retrieved from RT2 mice were implanted into the anterior chamber of the eye (ACE) of host mice, where they engrafted on the iris, recruited blood vessels and showed continuous growth. Noninvasive confocal and two-photon laser-scanning microscopy through the transparent cornea facilitated high-resolution imaging of tumour growth and angiogenesis. RT2 tumours in the ACE expanded up to 8-fold in size and shared hallmarks with tumours developing in situ in the pancreas. Genetically encoded fluorescent reporters enabled high-resolution imaging of stromal cells and tumour cell migration. Sunitinib treatment impaired RT2 tumour angiogenesis and growth, while overexpression of the vascular endothelial growth factor (VEGF)-B increased tumour angiogenesis though tumour growth was impaired. In conclusion, we present a novel platform for intravital high-resolution and 3D imaging of PNET biology and cancer drug assessment.


2020 ◽  
Vol 10 (7) ◽  
pp. 2502 ◽  
Author(s):  
Wei Liu ◽  
Qian Cheng ◽  
Linong Liu ◽  
Yun Wang ◽  
Jianfeng Zhang

The emerging applications of deep learning in solving geophysical problems have attracted increasing attention. In particular, it is of significance to enhance the computational efficiency of the computationally intensive geophysical algorithms. In this paper, we accelerate deabsorption prestack time migration (QPSTM), which can yield higher-resolution seismic imaging by compensating absorption and correcting dispersion through deep learning. This is implemented by training a neural network with pairs of small-sized patches of the stacked migrated results obtained by conventional PSTM and deabsorption QPSTM and then yielding the high-resolution imaging volume by prediction with the migrated results of conventional PSTM. We use an encoder-decoder network to highlight the features related to high-resolution migrated results in a high-order dimension space. The training data set of small-sized patches not only reduces the required high-resolution migrated result (for instance, only several inline is required) but leads to a fast convergence in training. The proposed deep-learning approach accelerates the high-resolution imaging by more than 100 times. Field data is used to demonstrate the effectiveness of the proposed method.


Sign in / Sign up

Export Citation Format

Share Document