Data collection:

We grew black gram (Vigna mungo) plants under nutrient control environments. Firstly, black gram seeds had been planted in sand and grown for two weeks. After that, only healthy black gram seedlings with approximately the same height were chosen. Each was moved to a bottle of nutrient-controlled solution (soiless growing), and was then grown for 28 days. Seven types of solutions were used:

1. Complete (COM) nutrients

2. Calcium (Ca) deficient

3. Iron (Fe) deficient

4. Magnesium (Mg) deficient

5. Nitrogen (N) deficient

6. Phosphorus (P) deficient

7. Potassium (K) deficient

The solution was changed every week to maintain the amount of nutrients. For each type of solution, we grew 10-12 black gram plants; 8-10 plants for training and 2 plants for testing. For each plant, we took one or two leaf images (a younger leaf and an older leaf) every day (except day 20 due to a problem) under a controlled lighting condition. A DSLR camera, i.e., Canon EOS 550D with 18-55 mm and f/3.5-5.6 lens, was used for image acquisition. Each image was resized to 1296 x 864 pixels. We also manually segmented the target leaf in each image.

References:

• Kadipa Aung Myo Han and Ukrit Watchareeruetai, "Black gram plant nutrient deficiency classification in combined images using convolutional neural network," The 2020 International Electrical Engineering Congress (iEECON 2020), Chiang Mai, Thailand, March 4-6, 2020. DOI: 10.1109/iEECON48109.2020.229562

• Kadipa Aung Myo Han and Ukrit Watchareeruetai, "Classification of nutrient deficiency in black gram using deep convolutional neural networks," The 16th International Joint Conference on Computer Science and Software Engineering (JCSSE 2019), Pattaya, Thailand, July 10-12, 2019. DOI: 10.1109/JCSSE.2019.8864224

• Ukrit Watchareeruetai, Pavit Noinongyao, Chaiwat Wattanapaiboonsuk, Puriwat Khantiviriya, and Sutsawat Duangsrisai, "Identification of plant nutrient deficiencies using convolutional neural networks," The 6th International Electrical Engineering Congress (iEECON 2018), pp.679-682, Krabi, Thailand, March 7-9, 2018. DOI: 10.1109/IEECON.2018.8712217

Data collection:

The image data was collected in the same way as in the Plant Nutrient Deficiency Identification dataset. However, we also provide segmentation annotation of abnormal regions caused by nutrient deficiencies. The segmentation annotation was done manually by human. Five types of nutrient deficiencies (Ca, Fe, K, Mg, and N) and a controlled group (complete nutrition) were considered. We chose images of leaves on day 15 after nutrient deficiency stress began. The dataset consists of 90 images; 60 images of complete nutrition and 30 images of nutrient deficient leaves (6 images for each deficiency type).

References:

• Pavit Noinongyao, Ukrit Watchareeruetai, Puriwat Khantiviriya, Chaiwat Wattanapaiboonsuk, and Sutsawat Duangsrisai, "Separation of abnormal regions on black gram leaves using image analysis," The 14th International Joint Conference on Computer Science and Software Engineering (JCSSE 2017), pp.1-5, Nakhon Si Thammarat, Thailand, July 12-14, 2017. DOI: 10.1109/JCSSE.2017.8025951