Dataset of vector mosquito images

Mosquitoes pose substantial threat to public health resulting in million number of deaths wordlwide every year. They act as the vectors responsible for diseases such as Dengue, Yellow fever,Chikungunya, Zika etc. The harmful mosquito species are contained in the genera Aedes, Anopheles and Culex. Automated species identification of vectors is essential to implement targeted vector control strategies. The objective of the proposed paper is to construct a novel dataset of images of dangerous mosquito species. We have prepared a dataset of images of adult mosquitoes belonging to three species: Aedes Aegypti, Anopheles stephensi and Culex quinquefasciatus stored in two folders. The first folder comprises of total 2640 augmented images of mosquitoes belonging to the three species. The second folder contains original images of the the three species. The dataset is valuable for training machine and deep learning models for automatic species classification.


Specifications
Computer Vision and Pattern Recognition, Machine Learning, Entomology and insect science. Specific subject area: Morphological classification of mosquito species. Type of data: Images of Mosquitoes How data points were acquired: The images were captured with a 48 Mpx One Plus mobile phone camera in the day light condition. Data format: Raw images in JPEG file format. Description of data collection: Photographs of fresh mosquito specimens were shot at day light using high resolution mobile phone rear camera. Data source location: All

Value of the Data
• The dataset provides images of three mosquito species: Aedes Aegypti, Anopheles stephensi and Culex quinquefasciatus. • The dataset can be used to train mosquito species classification and prediction models. The dataset can potentially benefit the society in controlling mosquito borne diseases. • The dataset can be used to train automated species classification models which is a vital contribution for vector control. • Automated genera and species identification can be efficient as compared to the laborious and time consuming task of manual species identification carried out by entemologists.

Data Description
Mosquitoes of genera Aedes, Anopheles and Culex are vectors responsible for spreading diseases such as Dengue, Yellow fever, Chikungunya, Zika etc. [1] . Mosquito vector surveillance is carried out by local government to monitor the mosquito population and the species predominant in a geographic area to implement effective mosquito vector control plans [2] . Automated species classification can be an important contribution to target harmful species. Image processing techinques with machine learning algorithms can be used to train machine learning models to classify and predict the genera or species. Availability of a quality data set is a prerequisite to train such deep learning models [3 , 4 , 5] .
There are datasets which include geographical density and distribution record of vector mosquito species [6 , 7] . There are images datasets available which contain images of female mosquitoes belonging to i) Aedes genera (Aedes aegypti and Aedes albopictus species), ii) Aedes and Culex genera (Aedes aegypti and Aedes albopictus and Culex quinquefasciatus species and iii) Aedes, Anopheles and Culex species. [8 , 9 , 10] .
The images in these datasets were acquired using microscope and a digital camera. In our work we have included the three important harmful vector species i.e., Aedes aegypti, Anopheles stephensi and Culex quinquefasciatus of both sexes. Also, the images are captured with mobile phone camera.
The proposed dataset folder comprises of two folders. The folder named "Mosquito Images Original" contains original images of the three species stored under three sub folders that are created for each of the three species: Aedes Aegypti, Anopheles Stephensi and Culex Quinquefasciatus. The folder named "Mosquito Images Augmented" consists of total 2640 augmented images of the three species stored under the 3 corresponding subfolders. The pictures were captured with One Plus mobile 48 Mpx camera and were saved in JPEG file format. The original pictures are RGB images with dimension 30 0 0 x 40 0 0 pixels and 72 dpi. The augmented images are of resolution 256 × 256 pixels with 96 dpi. Table 1 presents the number of images and a sample picture of mosquito belonging to each of the three species..

Experimental Design
The images were collected during April 2022, at the mosquito colony maintained by Ross Life lab, Pune city of Maharashtra, India. Fig. 1 shows the steps involved in dataset construction Fig. 1. Data acquisition process process. The fresh adult mosquito specimens were imaged using a handheld smartphone camera. The species included were Aedes Aegypti, Anopheles Stephensi and Culex Quinquefasciatus. The species of the specimen were confirmed by an entomologist from the Ross Life.

Materials or Specification of Image Acquisition System
The imaging system consisted of a 48 Mpx One Plus mobile RGB camera. ( Table. 2 )

Ethics Statement
This data is available in the public domain, and no funding is received for the present effort. There is no conflict of interest.