librahfacebook/pededs

Pedestrian detection and Segmentation

Pedestrian detection and Segmentation

---
title: PDS: 行人检测及分割
author: librah
date: 2020/5/5
---

Introduction

Object Detection is a common Computer Vision problem which deals with identifying and locating object of certain classes in the image. Interpreting the object localization in various ways, including creating a bounding box around the object or marking every pixel in the image which contains the object (called segmentation).

Image segmentation creates a pixel-wise mask for each object in the image. This technique gives us a far more granular understanding of the object(s) in the image.

Models

The model is Mask R-CNN, which is based on top of Faster R-CNN. Mask R-CNN adds an extra branch into Faster R-CNN, which also predicts segmentation masks for each instance.

Structure

networks: Mask R-CNN, based on torchvision.models.detection. We use the FastRCNNPredictor to detection boxes and scores of objects, and use the MaskRCNNPredictor to object segmentation.

utils: some official application tools.

dataloader.py: define a custom dataset for Pedestrian.

experiments.py: train model, predict and show the results.

Results

Reference

[1] https://pytorch.org/tutorials/intermediate/torchvision_tutorial.html

[2] https://matplotlib.org/3.1.1/contents.html