Abstract Details
| Name: Kumar Pranshu Affiliation: ARIES, Nainital Conference ID: ASI2024_620 Title : Automated Classification of Transient Alerts in the context of the 4-m International Liquid Mirror Telescope Authors and Co-Authors : Kumar Pranshu, Kuntal Misra, Bhavya Ailawadhi, Monalisa Dubey, Naveen Dukiya, Brajesh Kumar, Vibhore Negi, Paul Hickson, Jean Surdej Abstract Type : Poster Abstract Category : Facilities, Technologies and Data science Abstract : The 4-m International Liquid Mirror Telescope is the first telescope dedicated to optical survey in India. Since it will acquire images of common LST fields across several nights, it makes it a very promising instrument to find several variable and transient sources by image subtraction technique. A plenty of sources found using this method turned out to be known asteroids and variable stars. Given this relatively large number of candidates, it becomes imperative to develop some techniques to classify these candidates. A CNN based candidate classifier module called NovaNet was developed that classifies these candidates into three categories based on the morphology of the 'host' in the reference image viz. 'extended-host', 'point-host' and 'host-less'. The classifier also flags the known asteroids among the candidates which can be rejected at a later stage. The remaining candidates can then be followed up according to the class priority decided by the user (Eg: Extended-host might be a likely extragalactic candidate as the host is likely to be a galaxy). The module has been integrated into the ILMT transient detection pipeline. |