search

yeephycho / tensorflow-face-detection

A mobilenet SSD based face detector, powered by tensorflow object detection api, trained by WIDERFACE dataset.
Apache License 2.0
768 stars 264 forks source link

Which pre-trained model is used for training? #64

Open Varat7v2 opened 4 years ago

Varat7v2 commented 4 years ago

Which model you used for training? Is it ssd mobilenet v1 coco with input size 300x300?

I think you had trained with input image size 512x512 right? I am also trying to train mobilenet V2 coco with input image size 512x512, but there seems to be no changes as compared to 300x300. How can I do that? I am using per-trained weights from Tensorflow zoo for fine-tuning.