{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T13:05:00Z","timestamp":1789391100719,"version":"build-2803163510"},"reference-count":28,"publisher":"MDPI AG","issue":"18","license":[{"start":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T00:00:00Z","timestamp":1789344000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62466013"],"award-info":[{"award-number":["62466013"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"award":["62466013"],"award-info":[{"award-number":["62466013"]}],"id":[{"id":"https:\/\/ror.org\/01h0zpd94","id-type":"ROR","asserted-by":"publisher"}]},{"name":"Guangxi Science and Technology Program","award":["Guike GXYD2504850002"],"award-info":[{"award-number":["Guike GXYD2504850002"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>The persistent challenge in visual object tracking, particularly for small and fast-moving targets, lies in the trade-off between effective resolution and contextual information. Fixed search regions cannot adapt to variations in target scale and motion, often resulting in degraded target representation and tracking failures. In this paper, we introduce AdaSAM2, a framework that formulates adaptive search-region selection as a sequential decision-making problem. Unlike conventional per-frame heuristics, our approach employs an event-driven reinforcement learning policy that selects the cropping scale only during initialization and unreliable tracking states, while reusing the previous configuration during stable tracking. To handle target disappearance, we further introduce a lost-aware recovery mechanism that combines progressive search-region enlargement with constrained policy re-selection. Extensive experiments on TSFMO, LaTOT, UAV123, and UAVDT demonstrate consistent improvements over the SAMITE baseline. For example, AdaSAM2 improves Success and Precision on TSFMO from 42.6% and 74.0% to 43.9% and 75.3%, respectively, while improving UAV123 Success and Precision from 69.6% and 92.7% to 71.1% and 94.8%. Moreover, the RL policy is activated on only 0.74% of processed frames on TSFMO, resulting in an average overhead of only 0.0085 ms per frame. These results demonstrate that adaptive input-space optimization can improve tracking accuracy while introducing negligible computational overhead.<\/jats:p>","DOI":"10.3390\/s26185819","type":"journal-article","created":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T12:09:51Z","timestamp":1789387791000},"page":"5819","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Learning Adaptive Search with Reinforcement Learning for Small and Fast Object Tracking"],"prefix":"10.3390","volume":"26","author":[{"given":"Binrui","family":"Liu","sequence":"first","affiliation":[{"name":"College of Computer Science and Engineering, Guilin University of Technology, Guilin 541006, China"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"conceptualization","vocabulary":"credit"},{"role":"investigation","vocabulary":"credit"},{"role":"visualization","vocabulary":"credit"},{"role":"methodology","vocabulary":"credit"},{"role":"software","vocabulary":"credit"},{"role":"data-curation","vocabulary":"credit"},{"role":"writing-original-draft","vocabulary":"credit"},{"role":"writing-review-editing","vocabulary":"credit"},{"role":"formal-analysis","vocabulary":"credit"},{"role":"validation","vocabulary":"credit"}]},{"given":"Xinyi","family":"Bo","sequence":"additional","affiliation":[{"name":"College of Computer Science and Engineering, Guilin University of Technology, Guilin 541006, China"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"writing-review-editing","vocabulary":"credit"},{"role":"methodology","vocabulary":"credit"},{"role":"validation","vocabulary":"credit"}]},{"given":"Wenbin","family":"Luo","sequence":"additional","affiliation":[{"name":"Guangxi Science and Technology Information Network Center, Nanning 530022, China"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"methodology","vocabulary":"credit"},{"role":"software","vocabulary":"credit"},{"role":"validation","vocabulary":"credit"}]},{"given":"Haolun","family":"Li","sequence":"additional","affiliation":[{"name":"College of Computer Science and Engineering, Guilin University of Technology, Guilin 541006, China"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"validation","vocabulary":"credit"}]},{"given":"Jingqi","family":"Wang","sequence":"additional","affiliation":[{"name":"College of Computer Science and Engineering, Guilin University of Technology, Guilin 541006, China"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"validation","vocabulary":"credit"}]},{"given":"Ge","family":"Zheng","sequence":"additional","affiliation":[{"name":"College of Computer Science and Engineering, Guilin University of Technology, Guilin 541006, China"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"validation","vocabulary":"credit"},{"role":"writing-review-editing","vocabulary":"credit"}]},{"given":"Shuiwang","family":"Li","sequence":"additional","affiliation":[{"name":"College of Computer Science and Engineering, Guilin University of Technology, Guilin 541006, China"}],"role":[{"vocabulary":"crossref","role":"author"},{"role":"writing-review-editing","vocabulary":"credit"},{"role":"funding-acquisition","vocabulary":"credit"},{"role":"project-administration","vocabulary":"credit"},{"role":"supervision","vocabulary":"credit"},{"role":"conceptualization","vocabulary":"credit"},{"role":"methodology","vocabulary":"credit"}]}],"member":"1968","published-online":{"date-parts":[[2026,9,14]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","unstructured":"Mueller, M., Smith, N., and Ghanem, B. (2016). A Benchmark and Simulator for UAV Tracking. Proceedings of the European Conference on Computer Vision (ECCV), Springer.","DOI":"10.1007\/978-3-319-46448-0_27"},{"key":"ref_2","doi-asserted-by":"crossref","unstructured":"Du, D., Qi, Y., Yu, H., Yang, Y., Duan, K., Li, G., Zhang, W., Huang, Q., and Tian, Q. (2018, January 8\u201314). The unmanned aerial vehicle benchmark: Object detection and tracking. Proceedings of the European Conference on Computer Vision (ECCV), Munich, Germany.","DOI":"10.1007\/978-3-030-01249-6_23"},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"Wu, Y., Wang, X., Yang, X., Liu, M., Zeng, D., Ye, H., and Li, S. (2025, January 11\u201315). Learning occlusion-robust vision transformers for real-time uav tracking. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, Nashville, TN, USA.","DOI":"10.1109\/CVPR52734.2025.01594"},{"key":"ref_4","doi-asserted-by":"crossref","unstructured":"Xue, C., Zhong, B., Liang, Q., Zheng, Y., Li, N., Xue, Y., and Song, S. (2025, January 10\u201317). Similarity-guided layer-adaptive vision transformer for UAV tracking. Proceedings of the Computer Vision and Pattern Recognition Conference, Nashville, TN, USA.","DOI":"10.1109\/CVPR52734.2025.00631"},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"2403","DOI":"10.1109\/TCSVT.2025.3599856","article-title":"Learning an adaptive and view-invariant vision transformer for real-time UAV tracking","volume":"36","author":"Wu","year":"2025","journal-title":"IEEE Trans. Circuits Syst. Video Technol."},{"key":"ref_6","doi-asserted-by":"crossref","unstructured":"Zheng, Y., Zhong, B., Liang, Q., Mo, Z., Zhang, S., and Li, X. (2024, January 20\u201327). Odtrack: Online dense temporal token learning for visual tracking. Proceedings of the AAAI Conference on Artificial Intelligence, Vancouver, BC, Canada.","DOI":"10.1609\/aaai.v38i7.28591"},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"18797","DOI":"10.1109\/JIOT.2026.3662378","article-title":"Smtrack: End-to-end trained spiking neural networks for multi-object tracking in rgb videos","volume":"13","author":"Zhong","year":"2026","journal-title":"IEEE Internet Things J."},{"key":"ref_8","doi-asserted-by":"crossref","unstructured":"Zhang, Z., Wu, F., Qiu, Y., Liang, J., and Li, S. (2022, January 4\u20138). Tracking small and fast moving objects: A benchmark. Proceedings of the Asian Conference on Computer Vision, Macao, China.","DOI":"10.1007\/978-3-031-26293-7_33"},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"10273","DOI":"10.1109\/TNNLS.2023.3239529","article-title":"Tiny object tracking: A large-scale dataset and a baseline","volume":"35","author":"Zhu","year":"2023","journal-title":"IEEE Trans. Neural Netw. Learn. Syst."},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Mirzaei, B., Nezamabadi-Pour, H., Raoof, A., and Derakhshani, R. (2023). Small object detection and tracking: A comprehensive review. Sensors, 23.","DOI":"10.3390\/s23156887"},{"key":"ref_11","unstructured":"Ravi, N., Gabeur, V., Hu, Y.T., Hu, R., Ryali, C., Ma, T., Khedr, H., R\u00e4dle, R., Rolland, C., and Gustafson, L. (2024). Sam 2: Segment anything in images and videos. arXiv."},{"key":"ref_12","unstructured":"Yang, C., Huang, H., Chai, W., Jiang, Z., and Hwang, J. (2024). Samurai: Adapting segment anything model for zero-shot visual tracking with motion-aware memory. arXiv."},{"key":"ref_13","unstructured":"Xu, Q., Zhu, L., Liu, C., Lin, G., Long, C., Li, Z., and Zhao, R. (2025). Samite: Position prompted sam2 with calibrated memory for visual object tracking. arXiv."},{"key":"ref_14","doi-asserted-by":"crossref","unstructured":"Van Hasselt, H., Guez, A., and Silver, D. (2016, January 12\u201317). Deep reinforcement learning with double q-learning. Proceedings of the AAAI Conference on Artificial Intelligence, Phoenix, AZ, USA.","DOI":"10.1609\/aaai.v30i1.10295"},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"529","DOI":"10.1038\/nature14236","article-title":"Human-level control through deep reinforcement learning","volume":"518","author":"Mnih","year":"2015","journal-title":"Nature"},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"Bertinetto, L., Valmadre, J., Henriques, J.F., Vedaldi, A., and Torr, P.H. (2016). Fully-convolutional siamese networks for object tracking. Proceedings of the European Conference on Computer Vision, Springer.","DOI":"10.1007\/978-3-319-48881-3_56"},{"key":"ref_17","doi-asserted-by":"crossref","unstructured":"Li, B., Wu, W., Wang, Q., Zhang, F., Xing, J., and Yan, J. (2019, January 15\u201320). Siamrpn++: Evolution of siamese visual tracking with very deep networks. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, CA, USA.","DOI":"10.1109\/CVPR.2019.00441"},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Cui, Y., Jiang, C., Wang, L., and Wu, G. (2022, January 18\u201324). Mixformer: End-to-end tracking with iterative mixed attention. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, LA, USA.","DOI":"10.1109\/CVPR52688.2022.01324"},{"key":"ref_19","doi-asserted-by":"crossref","unstructured":"Song, Z., Yu, J., Chen, Y.P.P., and Yang, W. (2022, January 18\u201324). Transformer tracking with cyclic shifting window attention. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, LA, USA.","DOI":"10.1109\/CVPR52688.2022.00859"},{"key":"ref_20","doi-asserted-by":"crossref","unstructured":"Lin, T.Y., Doll\u00e1r, P., Girshick, R., He, K., Hariharan, B., and Belongie, S. (2017, January 21\u201326). Feature pyramid networks for object detection. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA.","DOI":"10.1109\/CVPR.2017.106"},{"key":"ref_21","doi-asserted-by":"crossref","unstructured":"Zhu, Z., Wang, Q., Li, B., Wu, W., Yan, J., and Hu, W. (2018, January 8\u201314). Distractor-aware siamese networks for visual object tracking. Proceedings of the European Conference on Computer Vision (ECCV), Munich, Germany.","DOI":"10.1007\/978-3-030-01240-3_7"},{"key":"ref_22","doi-asserted-by":"crossref","unstructured":"Kirillov, A., Mintun, E., Ravi, N., Mao, H., Rolland, C., Gustafson, L., Xiao, T., Whitehead, S., Berg, A.C., and Lo, W.Y. (2023, January 1\u20136). Segment anything. Proceedings of the IEEE\/CVF International Conference on Computer Vision, Paris, France.","DOI":"10.1109\/ICCV51070.2023.00371"},{"key":"ref_23","doi-asserted-by":"crossref","unstructured":"Videnovic, J., Lukezic, A., and Kristan, M. (2025, January 10\u201317). A distractor-aware memory for visual object tracking with sam2. Proceedings of the Computer Vision and Pattern Recognition Conference, Nashville, TN, USA.","DOI":"10.1109\/CVPR52734.2025.02259"},{"key":"ref_24","unstructured":"Schulman, J., Wolski, F., Dhariwal, P., Radford, A., and Klimov, O. (2017). Proximal policy optimization algorithms. arXiv."},{"key":"ref_25","doi-asserted-by":"crossref","unstructured":"Yun, S., Choi, J., Yoo, Y., Yun, K., and Young Choi, J. (2017, January 21\u201326). Action-decision networks for visual tracking with deep reinforcement learning. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA.","DOI":"10.1109\/CVPR.2017.148"},{"key":"ref_26","doi-asserted-by":"crossref","first-page":"10","DOI":"10.1016\/j.cviu.2018.05.009","article-title":"Real-time visual tracking by deep reinforced decision making","volume":"171","author":"Choi","year":"2018","journal-title":"Comput. Vis. Image Underst."},{"key":"ref_27","doi-asserted-by":"crossref","unstructured":"Huang, C., Lucey, S., and Ramanan, D. (2017, January 22\u201329). Learning policies for adaptive tracking with deep feature cascades. Proceedings of the IEEE International Conference on Computer Vision, Venice, Italy.","DOI":"10.1109\/ICCV.2017.21"},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"120880","DOI":"10.1109\/ACCESS.2021.3108623","article-title":"On the use of deep reinforcement learning for visual tracking: A survey","volume":"9","author":"Cruciata","year":"2021","journal-title":"IEEE Access"}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/26\/18\/5819\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,14]],"date-time":"2026-09-14T12:33:24Z","timestamp":1789389204000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/26\/18\/5819"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,14]]},"references-count":28,"journal-issue":{"issue":"18","published-online":{"date-parts":[[2026,9]]}},"alternative-id":["s26185819"],"URL":"https:\/\/doi.org\/10.3390\/s26185819","relation":{},"ISSN":["1424-8220"],"issn-type":[{"value":"1424-8220","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,14]]}}}