Skip to content

Chumsy0725/PASLCD

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 

Repository files navigation

Pose-Agnostic Scene-Level Change Detection (PASLCD)

The dataset and the code will be released here.

PASLCD Dataset

We introduce a new real-world dataset containing image sequences collected from ten challenging scenes captured using an iPhone following random and independent trajectories. Among the five indoor and five outdoor scenes, three from each are forward-facing and the remaining two are 360°. See below for details and visualizations:

Scene Indoor/Outdoor FF/360° Changes (click to expand)
Cantina Indoor FF Cantina
Lounge Indoor FF Lounge
Printing Area Indoor FF Printing Area
Lunch Room Indoor 360° Lunch Room
Meeting Room Indoor 360° Meeting Room
Garden Outdoor FF Garden
Pots Outdoor FF Pots
Zen Outdoor FF Zen
Playground Outdoor 360° Playground
Porch Outdoor 360° Porch

Each scene contains images from two reference (pre-change) instances — referred to as Instance 1 and Instance 2 in the paper and dataset — one inference (post-change) image set and labelled ground truth masks from the camera poses of the inference set. Each reference instance contains 50–109 images stored in <Scene>/<Instance_#>/input_orig. Each inference set contains 25 images stored in <Scene/<Instance_#>/input_orig_test and their respective masks in <Scene>/<Instance_#>/gt_mask. Between either reference instance and the inference set, there are 5–16 relevant structural and surface-level changes. These changes are listed in changes.txt under each <Scene>/<Instance_#> combination. Comparing Instance 1 with the inference set tests change detection under consistent lighting conditions, while comparing Instance 2 with the inference set tests change detection under different lighting conditions. See below for the directory structure:

├── /
│   ├── Cantina
│   │   ├── Instance_1
│   │   │   ├── changes.txt
│   │   │   ├── gt_mask
│   │   │   │   ├── Inst_1_test_IMG_nnnn.png
│   │   │   │   ├── ... 
│   │   │   ├── input_orig
│   │   │   │   ├── Inst_1_IMG_####.png
│   │   │   │   ├── ... 
│   │   │   ├── input_orig_test
│   │   │   │   ├── Inst_1_test_IMG_####.png
│   │   │   │   ├── ... 
│   │   ├── Instance_2
│   │   │   ├── changes.txt
│   │   │   ├── gt_mask
│   │   │   │   ├── Inst_2_test_IMG_nnnn.png
│   │   │   │   ├── ... 
│   │   │   ├── input_orig
│   │   │   │   ├── Inst_2_IMG_####.png
│   │   │   │   ├── ... 
│   │   │   ├── input_orig_test
│   │   │   │   ├── Inst_2_test_IMG_####.png
│   │   │   │   ├── ... 
│   ├── Lounge
│   ├── ...

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published