SynthDerm is a synthetically generated dataset inspired by the real-world characteristics of melanoma skin lesions in dermatology settings. These characteristics include whether the lesion is asymmetrical, its border is irregular or jagged, is unevenly colored, has a diameter more than 0.25 inches, or is evolving in size, shape, or color over time. These qualities are usually referred to as ABCDE of melanoma. We generate SynthDerm algorithmically by varying several factors: skin tone, lesion shape, lesion size, lesion location (vertical and horizontal), and whether there are surgical markings present. We randomly assign one of the following to the lesion shape: round, asymmetrical, with jagged borders, or multi-colored (two different shades of colors overlaid with salt-and-pepper noise). For skin tone values, we simulate Fitzpatrick ratings. Fitzpatrick scale is a commonly used approach to classify the skin by its reaction to sunlight exposure modulated by the density of melanin pigments in the skin. This rating has six values, where 1 represents skin that always burns (lowest melanin) and 6 represents skin that never burns in sunlight (highest melanin). For our synthetic generation, we consider six base skin tones that similarly resemble different amounts of melanin. We also add a small amount of random noise to the base color to add further variety. Overall, SynthDerm includes more than 2,600 images of size 64x64.