Deep Learning Methods for Improving Pollen Monitoring

Abstract

The risk of pollen-induced allergies can be determined and predicted based on data derived from pollen monitoring. Hirst-type samplers are sensors that allow airborne pollen grains to be detected and their number to be determined. Airborne pollen grains are deposited on adhesive-coated tape, and slides are then prepared, which require further analysis by specialized personnel. Deep learning can be used to recognize pollen taxa based on microscopic images. This paper presents a method for recognizing a taxon based on microscopic images of pollen grains, allowing the pollen monitoring process to be automated. In this research, a deep CNN (convolutional neural network) model was built from scratch. Publicly available deep neural network models, pre-trained on image data (not including microscopic pictures), were also used. The results show that even a simple deep learning model produces quite good results when the classification of pollen grain taxa is performed directly from the images. The best deep learning model achieved 97.88% accuracy in the difficult task of recognizing three types of pollen grains (birch, alder, and hazel) with similar structures. The derived models can be used to build a system to support pollen monitoring experts in their work.

Detection and Recognition of Pollen Grains in Multilabel Microscopic Images

“…This study extends the research presented in [ 11 ], in which image classification was performed. This task involves defining target classes and training a model to recognize them.…”

“…This study is a continuation of our previous work that attempted to create classification models based on deep learning [ 11 ]. The previous study was focused only on the identification of a taxon, which is crucial in pollen monitoring.…”

“…The neural network output contains class probabilities, almost equal to one for the classes detected in the image and close to zero for the other classes. The database used in [ 11 ] consisted of images containing exactly one object belonging to one class.…”

Field Evaluation of an Automated Pollen Sensor

“…The weak correlation results of weed and grass pollen indicate that the APS-300 system needs to strengthen its image library of these taxa to better train the pollen identification algorithm. Several previous studies also compared automatic pollen monitors with a Hirst type sampler, another NAB certified sampling method [ 14 , 19 , 24 ]. For example, Oteros et al evaluated a BAA500 automatic pollen sensor which also uses imaging recognition technology.…”

Analytics and Applications of Audio and Image Sensing Techniques

“…In [ 10 ], E. Kubera, A. Kubik-Komar, K. Piotrowska-Weryszko, and M. Skrzypiec applied deep learning in an image recognition task, for the automatic classification of pollen grains into 3 classes. Pollen monitoring is most commonly based on counting pollen grains (for each species) found on adhesive tape in pollen traps.…”