Transfer Learning of SSD-MOBILENET for a Custom Object

Iam using jetson-inference to achieve object detection on Xavier NX and the basic setup has worked well. Further, I tried training ssd-mobilenet for a custom object (cup) as per the documentation. But, the model is detecting other objects as trained object like, for example, the model is detecting person,pen,etc… as cup.

Eagerly, will be looking for the response.

Regards,
Manikandan K.