iVS-Dataset Statement

The dataset annotates the four commonly seen object types into four different categories of objects namely:

1.scooter
2.vehicle
3.pedestrian
4.bicycle

The iVS dataset consists of 89,002 annotated training images and 6,400 annotated testing images of resolution 1920x1080
目前本團隊已經公開89002張訓練資料與6400測試資料台灣
地區交通場域之行人、機車、自行車與汽車之標記資料庫，
免費提供學術界與產業界人士使用，進一步加速台灣地區深
度學習ADAS技術開發時程。
標註資訊包括物件類別與物件位置框。

Download link
Information doc

iVS-ODSEG-Dataset Statement

We provide 36,000 annotated 1280x720 training images, which part of them are revised from the original BDD100K semantic segmentation dataset with additional object bounding box and classes labels and part of them are our custom semantic segmentation dataset with additional object bounding box and classes labels.

1.background
2.main_lane
3.alter_lane
4.double_line
5.dashed_line
6.single_line

Download link