ResistorNet-Dataset

By Karabuk University Artificial Intelligence and Deep Learning Laboratory

Using CNN-InceptionV4 we reached 82% accuracy. 29 Epochs. Our paper can be found here.

Dataset

There are 37 class 3025 images in ResistorNet-DirençNet.

• 10 1/4W: 70

• 100 R 1/4W: 116

• 10 R 1W: 62

• 10 R 2W: 98

• 11 M 1/2W: 72

• 150 R 1/4W: 73

• 150 R 1/8W: 71

• 15 R 1/4W: 115

• 180 K 1/2W: 98

• 1 K 1/4W: 81

• 1 K 2W: 62

• 1 M 1/4W: 80

• 2.2 K 1/4W: 62

• 20 K 1/4W: 57

• 220 K 1/4W: 48

• 220 R 2W: 90

• 22 R 1/4W: 37

• 24 K 1/2W: 90

• 270 K 1/4: 75

• 27 R 1W: 54

• 2 R 1W: 76

• 3.9 K 1/4W: 80

• 330 R 1/4W: 51

• 33 K 2W: 123

• 4.7 K 1/4W: 90

• 470 R 1/4W: 173

• 470 R 1W: 66

• 5.1 K 1/4W: 40

• 5.6 K/4W: 87

• 56 K 1W: 49

• 5.1 K 1/4W: 77

• 6.8 R 1/4W: 73

• 620 R 1/4W: 81

• 68 K 1W: 95

• 7.5 K 1/4W: 80

• 8.2 K 1/4W: 74

• 820 R 1/4W: 97

• 4700Mohm : 102

  If you are using the dataset, please give a citation of this repository. The dataset can be downloaded here.

Details of datasets:

• Image size: 700 x 700 pixels
• Color space: RGB
• Number of classes: 37
• Resolution : 72 DPI

The devices used were Samsung Note 5, Nikon D7000.

Project Executives:

• Prof.Dr. Raif BAYIR - Academic Advisor
• Eralp ÖZCAN
• İlker ÖNALAN and Members of Artificial Intelligence and Deep Learning Lab.