Browsing by Author "Madushika, M.K.S."

Now showing 1 - 6 of 6
  • No Thumbnail Available
    Item
    Constraints Faced by Potato Farmers in Major Potato Growing Areas in Sri Lanka: An ICT based Intervention
    (Uva Wellassa University of Sri Lanka, 2020) Mohamed, M.S. A.; Wathugala, D.L.; Indika, W.A.; Madushika, M.K.S.; Piyaratne, M.K.D.K.; Samaraweera, G.C.
    Potato is one of the most attractive crops and plays an important role in human nutrition. However, the average yield of the potato is less than its yield potential and local production is insufficient to meet the demand in Sri Lanka. Therefore, a field survey was conducted in major potato growing districts such as Badulla, Nuwara-Eliya, and Jaffna to examine key challenges faced by potato farmers. One hundred potato farmers were selected through a purposive sampling technique and interviewed using a structured questionnaire. The collected data were analyzed using SPSS. The majority of farmers from study areas cultivate potato for consumption other than seed purpose and enriched with smallholdings. Results revealed that low farmgate prices (37%), pest and disease attacks (32%), and high cost of farm inputs (20%) are the major challenges faced by the majority of farmers. Low farmgate price for potatoes is mainly due to the involvement of a higher number of middlemen, the majority of farmers cultivate potato at the same time and low quality. Potato early blight, late blight, and bacterial wilt were common diseases; mite and whitefly attacks were the common pest attacks faced by the majority of farmers. Furthermore, high input price constituted the cost of seed potato (63%), fertilizer and chemical cost (18%), machinery and labor cost (12%), and cost for irrigation (7%). Thus, the importation of seed potatoes was the key influential factor for the increase of input price. The study recommends improving better coordination among stakeholders in the industry and it is crucial to direct farmers to follow proper cultivation and harvesting techniques. Further, the study suggested that it is important to come up with a mobilebased solution mainly among farmers to access context-specific information promptly and easily; that could be resolved major challenges faced by potato farmers in Sri Lanka. Keywords: Key challenges, Potato farmers, Yield
  • No Thumbnail Available
    Item
    A Model for a Mobile Application to Support Agro-ecological Zones based Crop Selection in Sri Lanka
    (Uva Wellassa University of Sri Lanka, 2020) Mohamed, M.S. A.; Wathugala, D.L.; Indika, W.A.; Madushika, M.K.S.; Piyaratne, M.K.D.K.; Samaraweera, G.C.
    Agriculture is the key source of livelihood and economic support for the Sri Lankan population. The farmer is the salient stakeholder in agriculture and he has to decide the appropriate crops for cultivation in every season. This decision should be primarily based on market conditions including pricing but several other factors such as climatic conditions of the area, land suitability, irrigation facilities, etc. should also take into consideration. Thus, crop selection is a vital and critical decision that farmer has to get in the farming lifecycle and many farmers face problems in selecting the right crops at the right time to grow. Therefore, in this study, a digital platform has been created to provide crop suitability information based on the agro-ecological zones in Sri Lanka. Providing information regarding suitable crops according to agro-ecological zones in Sri Lanka is the correct intervention to facilitate farmers during the crop selection stage. Contextual data for crop selection were mainly gathered through primary and secondary sources. The study learned that the agro-ecological zones have been classified based on different climatic zones, annual rainfall, terrain characteristics, available major soil groups, and recommended crops for all agro-ecological zones. A crop selection model was designed and listed out suitable crops based on 46 agro-ecological zones in Sri Lanka. Further, this decision is depended on the influence factors such as major cultivation seasons, irrigation types, and farmer preferences, etc. The designed model has been provided through a mobile-based platform to the farmers. Then, they could easily find recommended crops and varieties suitable to their farms by asking users to add their Province, District, Divisional Secretariat, and Grama Niladhari, division. Thus, the model will be promised in supporting farmers to increase the profit and social status of the farmers in Sri Lanka. Keywords: Agriculture, Agro-ecological zones, Crop selection, Farmers, Mobile application
  • No Thumbnail Available
    Item
    Music Emotion Recognition using Deep Neural Networks
    (Uva Wellassa University of Sri Lanka, 2021) Bathigama, T.H.; Madushika, M.K.S.
    Emotion is an integral part of music and a complex aspect of music that is not easily understood by machines. The emotional aspect of music is further complicated by the fact that it is a subjective experience that cannot be easily conveyed to machine. Although it is a complex problem, some progress has been made in this area suggesting that it might be feasible to develop computation models that can be used in real-world applications. Real-world applications of music emotion recognition systems range from entertainment to healthcare. In this paper we introduce a deep learning model that recognizes emotion in music from the audio signal. 1d and 2d convolution layers with different kernel sizes have been tested. Adaptive pooling layers have also been used to extract a fixed feature representation for the dense layers. We have also used trainable spectrogram extractors to learn different representations of the audio. To address the lack of data for the task of music emotion recognition we have also used the latest trends in audio data augmentation and converted it for music data. Till now we have been able to achieve an accuracy of about 0.92 for the PMEmo dataset and about 0.6 F-1 score from using the raw audio signal and 1D convolution layers to extract features. Preliminary experiments show that using 1d convolutions with the combination of learnable spectrograms performs satisfactorily. Further experiments are to be conducted using different combinations of raw audio and calculated features. Different model architectures using recurrent networks are also to be tested considering that audio has temporal relationship between each unit of time. Finally, the work done in this study is mainly to explore the high dimensional feature space of raw audio to extract features which can contribute to the recognition of emotion in music using automated methods such as convolution and recurrent layers. Keywords: Music Emotion Recognition; Deep Neural Networks; Music Data Augmentation; Arousal and Valence Prediction
  • No Thumbnail Available
    Item
    Present Situation of Intercropping in Potato Cultivation in Nuwara Eliya and Badulla Areas
    (Uva Wellassa University of Sri Lanka, 2020) Shanadi, A.T.; Samaraweera, G.C.; Wathugala, D.L.; Indika, W.A.; Madushika, M.K.S.; Piyaratne, M.K.D.K.
    Intercropping is one of the methods of increasing crop productivity. It has proved to achieve many advantages. However, with the increasing demand for local potatoes, intercropping with potato cultivation in Sri Lanka is still an open research challenge which is needed to be addressed. On the other hand, Potato is extensively cultivated in Nuwara Eliya and Badulla districts in Sri Lanka. Therefore, this study examined the present status of intercropping in potato cultivation in Nuwara Eliya and Badulla areas to determine the farmer’s knowledge and awareness on intercropping in potato cultivation. The study was based on primary data gathered through questionnaires from 100 potato farmers selected through the snowball technique. Out of 100 farmers, 48 farmers were selected in Nuwara Eliya district and others were selected in the Badulla district. The results obtained were analysed using descriptive statistics. Results indicated that 58% of farmers were aware of the intercropping. However, among them, due to the lack of knowledge and ignorance of the benefits of intercropping 30% of farmers were not practicing it. The other 28% of farmers did practice intercropping with short term vegetable crops such as beans, radish, and leaks, etc. They gained extra income from intercropping at a low cost of production by proper land use and maximum utilization of natural resources such as water and nutrient. The rest of the farmers (42%) were not aware of intercropping with potato and its benefits. Meanwhile, 54% of farmers claimed that they did not have vital information such as crop choices, cropping patterns, and amount of potential harvest, etc. Therefore, the study has brought out the urgency of the appropriate knowledge delivery method to disseminate the right knowledge on intercropping with potato and to create the awareness to encourage intercropping in potato cultivation which ensures food security, poverty reduction, and sustainable utilization of natural resources. Keywords: Intercropping, Food security, Potato cultivation
  • No Thumbnail Available
    Item
    Text-to-Face Generation with StyleGAN2
    (Uva Wellassa University of Sri Lanka, 2021) Ayanthi, D.M.A.; Madushika, M.K.S.
    Synthesizing images from a text description has become an active research area with the advent of Generative Adversarial Networks. It is a flexible way of generating images in a conditioned environment and has made significant progress in the recent years. The main goals of these models are to generate photo-realistic images that are well aligned with the input descriptions. Text-to-Face generation is a sub-domain of Text-to-Image generation that has been less explored because of its challenging nature. This is difficult because facial attributes are less specifically mentioned in descriptions and also because they are complex and has a wide variety. Although few works have been done in this domain, it has a variety of applications like in the fields of criminal investigation. But still there is the need to improve the image quality and how well the generated images match the input description. In this paper, we propose a novel framework for text-to-face generation using the state-of-the-art high-resolution image generator, StyleGAN2. For this task it is required to learn the mapping from the text space to the latent space of StyleGAN2. We chose BERT embeddings to encode the input descriptions. The text embedding mapped to the latent space, in turn was input to the StyleGAN2 model to generate facial images. We train and evaluate our model on the Text2Face dataset containing descriptions with at most 40 attributes for the images in the CelebA dataset. Our novel framework generates photo-realistic images by adopting StyleGAN2 and also improves the semantic alignment with the use of BERT embeddings that better capture the content of the description and the perceptual loss calculated using a pretrained VGG16 model. In the initial training we obtained a FID score of 370.57, Face Semantic Distance of 25.57 and a Face Semantic Similarity score of -0.002. With further training we believe the images could be made more realistic and semantically matching the input description. Keywords: Text-to-Image Synthesis; Text-to-Face Synthesis; StyleGAN2; High-resolution; Semantic Alignment
  • No Thumbnail Available
    Item
    User-Friendly Applications for Sri Lankan Farmers: “Govi Nena”
    (Uva Wellassa University of Sri Lanka, 2020) Gunawardana, D.A.Y.K.; Indika, W.A.; Madushika, M.K.S.; Wathugala, D.L.; Piyaratne, M.K.D.K.; Samaraweera, G.C.
    Most of the Sri Lankan farmers are used to cultivate any selected crop traditionally, as they are taught to be from their farming society. Due to this reason, there will be an overproduction from the same crop within a particular season. Even several crop forecasting applications are available in Sri Lanka, lack of reliability is the main drawback of these applications. These applications fail to guide the farmers to get a detailed review due to the interfaces of these applications are unable to visualize the required data reliably. In this research, a “Govi Nena” mobile-based application and a web-based dashboard were developed to select the most accurate crops to be cultivated to get the highest market demand and yield in an instant and reliable way with the userfriendly interfaces. Farmers have to enter the data about the conditions of their land, soil types according to their agro-ecological zone to the mobile application when they register to the application. The mobile application links with crop knowledgebase and provides crop lists of what they need to cultivate and crop calendar for each crop based on the planting dates inserted by the farmers. The analysed information based on the farmer inputs through a mobile application will be visualized on the dashboard, which consists of multifunctional, proper, understandable, and user-friendly interfaces including tables, charts, and graphs. When a farmer uses this dashboard, he/she can get a clear understanding of; how the yield of each crop varies with the time, most compatible crops which have been cultivated in different areas, how the market price for the crops varies with the time, etc. The dashboard shows the relationships, comparisons, composition, and distributions of the information/knowledge. Farmers can get a clear picture through understandable visualizations via this dashboard for selected crops. Keywords: Web-based dashboard, Mobile-based application, Sri Lankan agriculture, famers