Peeranat Fupongsiripan

Hi @villanuevab, did you have to modify some code after replacing nets/, preprocessing/ from SSD-Tensorflow repo? In vehicle-detection.ipynb for example, there are methods that do not exist in the SSD-Tensorflow...

Thanks for your response. So how did you manage to get SDC-Vehicle-Detection to work and get reasonable results and solve "TypeError: __init__() got multiple values for keyword argument 'dtype'" then?...

Hi @AloshkaD, Thanks for your reply. It helps me and my team alot. My team are taking Udacity's self driving car nano degree program and developing a self driving car...

Just finished watching it. Great video. Thank you and good luck with the interview :)

@AloshkaD Thanks for your reply. It is not too late :) I have trained my network with initialized weights from weights_SSD300.hdf5 and set to train for 200 epochs with a...

I can't find infrence.py in the repo. Are you sure it's from the same project? Btw, http://www.vision.caltech.edu/Image_Datasets/CaltechPedestrians/ seems promising for pedestrian dataset.

@AloshkaD Sorry for late reply. I've been implementing mAP and preparing pedestrian dataset as you suggested. Almost finish now. So to answer your questions. 1. I didn't include the dataset...

@AloshkaD thanks for reviewing my code. The gt_pascal.pkl was provided in the original repo(see https://github.com/rykov8/ssd_keras). It contains annotations for the other dataset. I didn't touch it. I just printed in...

@AloshkaD OK. I'm now re-training everything. This time learning details can be visualized with tensorboard and are evaluated every epoch with mAP, precision, recall, f1, and loss metrics. As you...

No. It took me quite some time so I gave up :D Right now, I only plot learning details.