Combining Deep Learning with Single-Spectrum UV Imaging for Rapid Detection of HNSs Spills

Deep or Shallow? A Comparative Analysis on the Oil Species Identification Based on Excitation-Emission Matrix and Multiple Machine Learning Algorithms

With the continuous expansion of petroleum extraction, transportation, and storage, the risk of oil spills at sea has also increased, posing a serious threat to marine safety. The excitation-emission matrix (EEM), which is composed of the fluorometric spectra under multiple excitation wavelengths, becomes a feasible approach to identify oil species. Despite the fact that various machine learning models have been applied to analyse EEMs of oil pollutants, it is unclear how much improvements the deep learning models have achieved, especially comparing with the shallow learning models. This paper presents a comparative analysis on the oil species identification using four types of machine learning models: random forest (RF), support vector machine (SVM), back propagation neural network (BPNN), and deep convolutional neural network (DCNN). The fluorescence of some common oils was excited using a tuneable xenon lamp and collected with a high-resolution spectrometer to form the EEMs for model training and testing.The results show that SVM, BPNN, and DCNN achieved high identification accuracies that are more than 93% on all types of oils tested in the study. The two deep learning models didn't have significant improvement over the SVM model. Considering the fact that the deep learning models require much larger number of calculations and longer running time, the SVM tends to be more suitable for oil species identification when considering the balance between the model accuracy and efficiency. This study provides some guidance on the choices of oil species identification model in the cases of oil spills.

Experimental Analysis on the Ultraviolet Imaging of Oil Film on Water Surface: Implication for the Optimal Band for Oil Film Detection Using Ultraviolet Imaging

Passive ultraviolet (UV) imaging is a potential way to detect thin oil film on water surface due to the high reflectivity of oil film under UV bands. This study conducted an outfield imaging experiment on oil film under the UV bands at 300 nm, 310 nm, and 330 nm. The obtained images were visually compared and quantitatively analyzed using imaging quality index (IQI). The experimental results indicated that the UV images obtained under 300 nm are not able to distinguish between oil film and clean water; those obtained under 310 nm achieve high contrast between oil film and clean water, but low average gray value; those obtained under 330 nm have high IQI and thus may be the optimal wavelength for UV imaging of oil film on water surface. This study provides a guidance to the choices of bands for the oil film detection using passive UV imaging method.

A Two-Stage Low-Altitude Remote Sensing Papaver Somniferum Image Detection System Based on YOLOv5s+DenseNet121

Papaver somniferum (opium poppy) is not only a source of raw material for the production of medical narcotic analgesics but also the major raw material for certain psychotropic drugs. Therefore, it is stipulated by law that the cultivation of Papaver somniferum must be authorized by the government under stringent supervision. In certain areas, unauthorized and illicit Papaver somniferum cultivation on private-owned lands occurs from time to time. These illegal Papaver somniferum cultivation sites are dispersedly-distributed and highly-concealed, therefore becoming a tough problem for government supervision. The low-altitude inspection of Papaver somniferum cultivation by unmanned aerial vehicles has the advantages of high efficiency and time saving, but the large amount of image data collected needs to be manually screened, which not only consumes a lot of manpower and material resources but also easily causes omissions. In response to the above problems, this paper proposed a two-stage (target detection and image classification) method for the detection of Papaver somniferum cultivation sites. In the first stage, the YOLOv5s algorithm was used to detect Papaver somniferum images for the purpose of identifying all the suspicious Papaver somniferum images from the original data. In the second stage, the DenseNet121 network was used to classify the detection results from the first stage, so as to exclude the targets other than Papaver somniferum and retain the images containing Papaver somniferum only. For the first stage, YOLOv5s achieved the best overall performance among mainstream target detection models, with a Precision of 97.7%, Recall of 94.9%, and mAP of 97.4%. For the second stage, DenseNet121 with pre-training achieved the best overall performance, with a classification accuracy of 97.33% and a Precision of 95.81%. The experimental comparison results between the one-stage method and the two-stage method suggest that the Recall of the two methods remained the same, but the two-stage method reduced the number of falsely detected images by 73.88%, which greatly reduces the workload for subsequent manual screening of remote sensing Papaver somniferum images. The achievement of this paper provides an effective technical means to solve the problem in the supervision of illicit Papaver somniferum cultivation.