Project Objectives:
Pest Identification and Control System deals with the modern-day problem in the agricultural sector. This problem arises in the form of pest infestation. Farmers are unable to identify these pests as they cannot afford to test these pest infected crops due to which a large portion of their crops gets ruined. Pest species can result in major infestation which can cause severe damage to agricultural land and will result in high financial loss for farmers. The first step to prevent crop damages is by identifying these pests, through methods that involve use of artificial intelligence and machine learning. Then the second step is to stop these pest attacks from occurring by using pesticides. Lastly the farmers should be provided with information regarding crops, crop seasons (Sowing and harvesting months) and pest that commonly attack these crops.
Socio-economic Benefits:
The method of crop pest detection is a challenging task for formers as a significant portion of the crops are damaged and the quality is dragged due to pest attack. Many of the rural areas lack the testing facilities and users cannot afford to test their crops due to which not only the users but the agriculture sector also suffers as well. Therefore, P.I.C will assist users in identifying the pest and suggest them a suitable pesticide to eliminate this threat as well. By using this application, the user can save their cost of testing of these pest and different pesticides, as these users can be spared traveling costs of going into different cities for testing and identification purpose. The application will also provide user with information about crop seasons along with crop information, sowing months and harvesting season and common pest that attack these crops. This information can assist farmers and prevent less drawbacks in the agriculture sector.
Project Methodology:
There have been some techniques used, like in deep learning (CNN), they have utilized 3-4 classes (9-24 images) from IP l02 dataset, and the system accuracy achieved is 89%. Other methods such as machine learning approaches (ANN and SVM) have also been used to identify the pest that harm the rice, by making use of 19 pest species these systems were able to obtain 92 % accuracy. These methods have made use of some classes due to which they still lack in accuracy. There is room for improvement. To address this issue, we have proposed a web-based application to identify the pest attack on crops. The system will utilize 8-10 classes from the dataset compared to the ones used in the systems explained above to further improve the accuracy results. The system will use deep learning algorithm such as CNN- (transfer learning) for identifying the pest. Furthermore, the proposed system will also provide the details of pest species and recommend suitable pesticide to avoid an infestation in the future. Users can also find the details of the crop season to reduce the risk of pest attacks. Urdu translated output will be provided in search results to make the information easier to access for any kind of user.
Project Outcome:
1. P.I.C will save the user all the hassle of physical testing and identification as it will provide pest information after identification as well.
2. In addition, P.I.C will also provide recommendation in form of pesticide to the user, so that user can make proactive plans to avoid pest attacks in the future.
3. Our system will not only aid user but can help in minimizing the losses suffered by the agriculture sector in form of pest attacks as well.
4. Our project will aid users by facilitating them with information in form of crop season, sowing months, harvesting months and Urdu translated search results as well.
