Name
Abstract | ||
|---|---|---|
In this notebook paper we describe our participation in the NIST TRECVID 2007 evaluation. We took part in two tasks of benchmark this year including high-level feature extraction and interactive search. For high-level feature extraction, we submitted 4 runs. FD_SVM: using SVM. FD_SVM_BN: using SVM and ontology. FD_MTL: using multi-task learning. FD_MTL_BN: using multi-task learning and ontology. Evaluation results illustrate that there are both advantages and disadvantages exist in all methods. For search, we submitted 5 interactive runs. Fudan_P: using multi-model and AP-based fusion. Fudan_R: using multi-model and MGR fusion. Fudan_C: cross system retrieval. Fudan_T: textual retrieval Fudan_I: image retrieval Evaluation results illustrate that the AP-based fusion method yields higher precision while the MGR fusion method finds more positive shots than other runs. We also experimented with simple cross system interactive retrieval to estimate the impact of manual browsing on the results. |
Year | Venue | Field |
|---|---|---|
2008 | TREC Video Retrieval Evaluation | Concept map,Visual search,Computer science,TRECVID,Full text search,Speech recognition,NIST |
DocType | Citations | PageRank |
Conference | 2 | 0.39 |
References | Authors | |
11 | 18 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Xiangyang Xue | 1 | 2466 | 154.25 |
| Hui Yu | 2 | 13 | 7.78 |
| Hong Lu | 3 | 43 | 8.61 |
| Yue-Fei Guo | 4 | 172 | 13.22 |
| Yuejie Zhang | 5 | 127 | 25.82 |
| Shile Zhang | 6 | 10 | 1.93 |
| Bin Li | 7 | 68 | 15.59 |
| Bolan Su | 8 | 472 | 22.37 |
| Yingbin Zheng | 9 | 191 | 16.70 |
| Wenjian Zhou | 10 | 4 | 1.12 |
| Lei Cen | 11 | 2 | 1.74 |
| Jie Zhang | 12 | 47 | 15.01 |
| Yu Jiang | 13 | 11 | 3.60 |
| Jiahui Qi | 14 | 2 | 0.39 |
| Jiaojiao Lu | 15 | 2 | 0.39 |
| Qian Diao | 16 | 2 | 0.39 |
| Zhenzhen Shi | 17 | 3 | 1.16 |
| Zichen Sun | 18 | 36 | 2.51 |