- คำอธิบาย :
ชุดข้อมูล Cars ประกอบด้วยรูปภาพ 16,185 รูปจากรถยนต์ 196 คลาส ข้อมูลถูกแบ่งออกเป็นรูปภาพการฝึก 8,144 รูป และรูปภาพทดสอบ 8,041 รูป โดยแต่ละคลาสจะถูกแบ่งประมาณ 50-50 รูป โดยทั่วไปชั้นเรียนจะอยู่ที่ระดับยี่ห้อ รุ่น ปี เช่น 2012 Tesla Model S หรือ 2012 BMW M3 coupe
เอกสารประกอบเพิ่มเติม : สำรวจในเอกสารด้วยรหัส
หน้าแรก : https://ai.stanford.edu/~jkrause/cars/car_dataset.html
ซอร์สโค้ด :
tfds.image_classification.Cars196รุ่น :
-
2.0.0: การเปิดตัวครั้งแรก -
2.0.1: อัปเดต URL ของเว็บไซต์ -
2.1.0(ค่าเริ่มต้น): แก้ไขข้อบกพร่อง https://github.com/tensorflow/datasets/issues/3927
-
ขนาดการดาวน์โหลด :
1.82 GiBขนาดชุดข้อมูล :
1.85 GiBแคชอัตโนมัติ ( เอกสาร ): No
แยก :
| แยก | ตัวอย่าง |
|---|---|
'test' | 8,041 |
'train' | 8,144 |
- โครงสร้างคุณสมบัติ :
FeaturesDict({
'bbox': BBoxFeature(shape=(4,), dtype=float32),
'id': Text(shape=(), dtype=string),
'image': Image(shape=(None, None, 3), dtype=uint8),
'label': ClassLabel(shape=(), dtype=int64, num_classes=196),
})
- เอกสารคุณสมบัติ :
| คุณสมบัติ | ระดับ | รูปร่าง | ประเภทD | คำอธิบาย |
|---|---|---|---|---|
| คุณสมบัติDict | ||||
| บีบ็อกซ์ | คุณสมบัติ BBox | (4,) | ลอย32 | |
| รหัส | ข้อความ | เชือก | ||
| ภาพ | ภาพ | (ไม่มี ไม่มี 3) | uint8 | |
| ฉลาก | ClassLabel | int64 |
คีย์ภายใต้การดูแล (ดู
as_superviseddoc ):('image', 'label')รูป ( tfds.show_examples ):

- ตัวอย่าง ( tfds.as_dataframe ):
- การอ้างอิง :
@inproceedings{KrauseStarkDengFei-Fei_3DRR2013,
title = {3D Object Representations for Fine-Grained Categorization},
booktitle = {4th International IEEE Workshop on 3D Representation and Recognition (3dRR-13)},
year = {2013},
address = {Sydney, Australia},
author = {Jonathan Krause and Michael Stark and Jia Deng and Li Fei-Fei}
}