IEEE - Trans. ITS
[
"Keywords" => [ "Vehicle part" , "Vehicle parts"],
"Year_Filter" => [
"2012-2016" => 14,
"2015-2016" => 6
]
],
[
"Keywords" => [ "Vehicle detection"],
"Year_Filter" => [
"2015-2017" => 8,
]
]
Highly Related Paper
1. [2014] Rear-View Vehicle Detection and Tracking by Combining Multiple Parts for Complex Urban Surveillance pdf
[
"Main" => "Rear Detection with Mulitparts",
"Features" =>
[
"target" => "license plate+rear lamps",
"detection_method" => "color conversion"
],
"Vehicle Detection" => "MRF to localize"
"Tracking" => "Kalman Filter"
]
PROS :

The relationships between the part nodes are represented by the Gaussian mixture model (GMM)
EXPERIMENT
[ "DB descript", "Accuracy", "Time" ]
2. [2013] Vehicle Detection by Independent Parts for Urban Driver Assistance pdf
Too Old
3. [2016] Multipart Vehicle Detection Using Symmetry-Derived Analysis and Active Learning pdf
multipart-based vehicle detection algorithm that employs Haar-like features and Adaboost classifiers for the detection of fully and partially visible rear views of vehicles.
4. [2016] Probabilistic Inference for Occluded and Multiview On-road Vehicle Detection pdf

Journal Construction
Selling point
- ( Wheel part based + Spatial Relation Model )
- Overtaking Vehicles
- Abstract
Introduction
- ITS => avoid accident
- vehicle accidnet => [ "對撞","側撞","同向擦撞","追撞" ]
- According to statistics from TW Gov., side crash 34% (?)
- "",
- Occlusion
- => unreasonable to treat the vehicle as a whole
- => a vehicle is composed of mulitparts
- => unreasonable to treat the vehicle as a whole
- Wheels => stable vehcile part which can represent vehicles which earliest appear in the view
- Wheel part based Vehicle Detection
- ITS => avoid accident
Related work 
  [Offline] ( HOG+MB-LBP ) + Adaboost
- 3.1 Vehicle Decomposed + 3 aspect ratio of wheels (落點分佈圖)
- 3.2 Detectors training process
- 3.3 Spatial Relation Model
[Online] Vehicle Detection
- 4.1 (ROI + Wheel Filtering) => Wheel Detection
- 4.2 Lane Detection + Localization
Experimental Result
5.1 Data set table 
5.2 Performance table (Vehicle) 
5.3 Performance table (Motor)
5.4 Performance table (Bike)
5.5 Time Consuming (???)
Conclusion
Reference
material
1.A1類:造成人員當場或二十四小時內死亡之交通事故。 2.A2類:造成人員受傷或超過二十四小時死亡之交通事故。 3.A3類:僅有財物損失之交通事故。
車與車-對撞 | 車與車-對向擦撞 | 車與車-同向擦撞 | 車與車-追撞 | 車與車-倒車撞 | 車與車-路口交岔撞 | 車與車-側撞 | 車與車-其他 | |
---|---|---|---|---|---|---|---|---|
102 | 3361 | 9243 | 34073 | 25857 | 2232 | 30713 | 93646 | 34645 |
103 | 3433 | 9682 | 36521 | 28730 | 2476 | 33381 | 102038 | 43367 |
104 | 3288 | 9583 | 34811 | 30509 | 2341 | 32147 | 100246 | 45474 |
Total | 10082 | 28508 | 105405 | 85096 | 7049 | 96241 | 295930 | 123486 |
Percentage | 1.3410535% | 3.79198108% | 14.0204071% | 11.319013% | 0.93762013% | 12.801461% | 39.3630195% | 16.4254446% |
751797 |
http://www.digitimes.com.tw/seminar/III_20161125/
而台灣與東南亞國家倚重2輪車為交通工具,根據台灣2013~2015交通事故統計資料,側撞(Side Crash)事故比例佔所有交通事故中的34.3%,側撞的車輛類型又以機車佔21.6%最高,小型車11.3%次之,而側撞事故發生地點絕大多數發生在十字路口上;有鑑於此,在台灣經濟部技術處支持下、財團法人資訊工業策進會智慧網通系統研究所(資策會智通所) 於今年六月與日本豐田IT開發中心合作,在「車聯網」及「自動駕駛」等相關議題上進行合作開發,希望解決社會大眾對於交通安全需求的車聯網應用服務。
Two-wheelers are one of the main transportation in Taiwan and Southeast Asian countries. According to Taiwan's 2013-2015 traffic accident statistics, side crash accounted for 34.3% of all traffic accidents. Side crash vehicle types: the motorcycle is the highest, accounting for 21.6%; the small car is the second, accounting for 11.3%. Side crash occurred mostly in the intersections, and we hope to improve traffic safety by this activity.
統計我國 103 年路口機車事故,發現其中有 46.7%是直行車輛 與其他左右轉汽機車碰撞造成的,此類肇事型態稱為側撞。以 103 年 為例,發生在路口的機車側撞車禍約 62,179 件,奪走 233 條機車族 生命,並造成 82,024 人
一般車輛肇事可分為對撞型態(碰撞角度介於180度與135度之間)事故、角撞型態(碰撞角度介於135度與45度之間)事故、側撞型態(碰撞角度介於45度與0度之間)事故及追撞型態(碰撞角度幾近於0度)事故[3]。