PyTorch实现 MobileNetV3-SSD 用于目标检测

flyfish

MobileNetV3-SSD implementation in PyTorch

环境
操作系统: Ubuntu18.04
Python: 3.6
PyTorch: 1.1.0

下载地址：
https://github.com/shaoshengsong/MobileNetV3-SSD

使用MobileNetV3-SSD实现目标检测
代码参考（严重参考以下代码）

一 SSD部分
https://github.com/amdegroot/ssd.pytorch
A PyTorch Implementation of Single Shot MultiBox Detector

二 MobileNetV3 部分

1 https://github.com/xiaolai-sqlai/mobilenetv3
mobilenetv3 with pytorch，provide pre-train model

2 https://github.com/kuan-wang/pytorch-mobilenet-v3
MobileNetV3 in pytorch and ImageNet pretrained models

3 https://github.com/leaderj1001/MobileNetV3-Pytorch
Implementing Searching for MobileNetV3 paper using Pytorch

4 https://github.com/qfgaohao/pytorch-ssd （严重抄袭此处）
MobileNetV1-SSD/SSD-lite，MobileNetV2-SSD/SSD-lite部分

MobileNetV1, MobileNetV2, VGG based SSD/SSD-lite implementation in Pytorch 1.0 / Pytorch 0.4. Out-of-box support for retraining on Open Images dataset. ONNX and Caffe2 support. Experiment Ideas like CoordConv.
no discernible latency cost.

针对4我这里没有做MobileNetV1, MobileNetV2等等代码兼容，只有MobileNetV3可用

下载数据
本例是以蛋糕和面包为例，原因是数据量小
所有类别总大小是561G，蛋糕和面包是3.2G

python3 open_images_downloader.py --root /media/santiago/a/data/open_images --class_names “Cake,Bread” --num_workers 20

训练过程
首次训练
python3 train_ssd.py --dataset_type open_images --datasets /media/santiago/data/open_images --net mb3-ssd-lite --scheduler cosine --lr 0.01 --t_max 100 --validation_epochs 5 --num_epochs 100 --base_net_lr 0.001 --batch_size 5

预加载之前训练的模型
python3 train_ssd.py --dataset_type open_images --datasets /media/santiago/data/open_images --net mb3-ssd-lite --pretrained_ssd models/mb3-ssd-lite-Epoch-99-Loss-2.5194434596402613.pth --scheduler cosine --lr 0.01 --t_max 100 --validation_epochs 5 --num_epochs 200 --base_net_lr 0.001 --batch_size 5

测试一张图片
python run_ssd_example.py mb3-ssd-lite models/mb3-ssd-lite-Epoch-99-Loss-2.5194434596402613.pth models/open-images-model-labels.txt /home/santiago/picture/test.jpg