Understanding population dynamics is critical for effective conservation efforts, particularly as environmental pressures such as habitat loss and climate change intensify. Traditional mark-recapture methods are valuable for estimating population size, survival rates, and movement patterns. However, they often rely on invasive techniques that may harm the animals being studied or affect their natural behaviors. To address this, non-invasive alternatives such as photo-recognition software have been developed, particularly for large mammals. More recent extensions of photo-recognition to smaller cryptic species have mostly been restricted to laboratory studies with captive animals. This study investigates the efficacy of photo-recognition on small cryptic species in a field setting. To do so we used the photo-recognition software Hotspotter and the lesser siren (Siren intermedia), a small benthic salamander, as our model. Sirens were captured across three human made ponds in the Davy Crockett National Forest. We assessed the ability of photo-recognition to identify individuals across time while using PIT-tags as redundant marks. I assessed the accuracy of HotSpotter through False Acceptance Rates (FAR) and False Rejection Rates (FRR). Overall low FAR (1.6%) and FRR (5.5%) indicated that photo-recognition was a viable tool for photo-recognition. FAR and FRR were spatially and temporally variable. Most false rejections occurred in a single pond and matching accuracy decreased over time. This study indicates photo recognition is a promising non-invasive tool for field-based individual recognition, but long-term studies may require a redundant identification method for species with ontogenetic shifts in patterns.