This is an unofficial inplementation of SSD: Single Shot MultiBox Detector in PyTorch. This project is based on the work here. Thanks to @amdegroot. Currently, only support PASCAL VOC 2007 dataset.
CUDA 10
python3.5+
Pytorch
(tested on 1.0.0)Visdom
(for visualization training process)
Run the command and it will download and unzip the dataset into './data'
folder.
$ bash scripts/voc2007.sh
- Download the fc-reduced VGG-16 PyTorch base network weights here:
$ mkdir weights
$ cd weights
$ wget https://s3.amazonaws.com/amdegroot-models/vgg16_reducedfc.pth
- Run the following script:
$ bash train.sh
- Download the PyTorch pre-trained SSD300 model on VOC from here:
$ cd weights
$ wget https://s3.amazonaws.com/amdegroot-models/ssd300_mAP_77.43_v2.pth
- Run the following script:
$ bash test.sh