Combining Domain Adaptation and Spatial Consistency for Unseen Fruits Counting: A Quasi-Unsupervised Approach

Combining Domain Adaptation and Spatial Consistency for Unseen Fruits Counting: A Quasi-Unsupervised Approach. Bellocchio, E., Costante, G., Cascianelli, S., Fravolini, M., L., & Valigi, P. IEEE Robotics and Automation Letters, 5(2):1079-1086, Institute of Electrical and Electronics Engineers Inc., 4, 2020.

Paper doi abstract bibtex

Autonomous robotic platforms can be effectively used to perform automatic fruits yield estimation. To this aim, robots need data-driven models that process image streams and count, even approximately, the number of fruits in an orchard. However, training such models following a supervised paradigm is expensive and unpractical. Extending pre-trained models to perform yield estimation for a completely new type of fruit is even more challenging, but interesting since this situation is typical in practice. In this work, we combine a State-of-the-Art weakly-supervised fruit counting model with an unsupervised style transfer method for addressing the task above. In this sense, our proposed approach is quasi-unsupervised. In particular, we use a Cycle-Generative Adversarial Network (C-GAN) to perform unsupervised domain adaptation and train it alongside with a Presence-Absence Classifier (PAC) that discriminates images containing fruits or not. The PAC produces the weak-supervision signal for the counting network, that can then be used on the target orchard directly. Experiments on datasets collected in four different orchards show that the proposed approach is more accurate than the supervised baseline methods.

@article{
 title = {Combining Domain Adaptation and Spatial Consistency for Unseen Fruits Counting: A Quasi-Unsupervised Approach},
 type = {article},
 year = {2020},
 keywords = {Agricultural automation,robotics in agriculture and forestry,visual learning},
 pages = {1079-1086},
 volume = {5},
 month = {4},
 publisher = {Institute of Electrical and Electronics Engineers Inc.},
 day = {1},
 id = {4cacbaff-57d3-31f6-a4da-40c99c38d3bb},
 created = {2023-01-06T11:09:15.016Z},
 accessed = {2023-01-06},
 file_attached = {true},
 profile_id = {235249c2-3ed4-314a-b309-b1ea0330f5d9},
 group_id = {5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1},
 last_modified = {2023-01-06T11:09:22.711Z},
 read = {false},
 starred = {false},
 authored = {false},
 confirmed = {false},
 hidden = {false},
 folder_uuids = {bd3c6f2e-3514-47cf-bc42-12db8b9abe45},
 private_publication = {false},
 abstract = {Autonomous robotic platforms can be effectively used to perform automatic fruits yield estimation. To this aim, robots need data-driven models that process image streams and count, even approximately, the number of fruits in an orchard. However, training such models following a supervised paradigm is expensive and unpractical. Extending pre-trained models to perform yield estimation for a completely new type of fruit is even more challenging, but interesting since this situation is typical in practice. In this work, we combine a State-of-the-Art weakly-supervised fruit counting model with an unsupervised style transfer method for addressing the task above. In this sense, our proposed approach is quasi-unsupervised. In particular, we use a Cycle-Generative Adversarial Network (C-GAN) to perform unsupervised domain adaptation and train it alongside with a Presence-Absence Classifier (PAC) that discriminates images containing fruits or not. The PAC produces the weak-supervision signal for the counting network, that can then be used on the target orchard directly. Experiments on datasets collected in four different orchards show that the proposed approach is more accurate than the supervised baseline methods.},
 bibtype = {article},
 author = {Bellocchio, Enrico and Costante, Gabriele and Cascianelli, Silvia and Fravolini, Mario Luca and Valigi, Paolo},
 doi = {10.1109/LRA.2020.2966398},
 journal = {IEEE Robotics and Automation Letters},
 number = {2}
}

Downloads: 0

{"_id":"ausfpfHLhEmtLKwM4","bibbaseid":"bellocchio-costante-cascianelli-fravolini-valigi-combiningdomainadaptationandspatialconsistencyforunseenfruitscountingaquasiunsupervisedapproach-2020","author_short":["Bellocchio, E.","Costante, G.","Cascianelli, S.","Fravolini, M., L.","Valigi, P."],"bibdata":{"title":"Combining Domain Adaptation and Spatial Consistency for Unseen Fruits Counting: A Quasi-Unsupervised Approach","type":"article","year":"2020","keywords":"Agricultural automation,robotics in agriculture and forestry,visual learning","pages":"1079-1086","volume":"5","month":"4","publisher":"Institute of Electrical and Electronics Engineers Inc.","day":"1","id":"4cacbaff-57d3-31f6-a4da-40c99c38d3bb","created":"2023-01-06T11:09:15.016Z","accessed":"2023-01-06","file_attached":"true","profile_id":"235249c2-3ed4-314a-b309-b1ea0330f5d9","group_id":"5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1","last_modified":"2023-01-06T11:09:22.711Z","read":false,"starred":false,"authored":false,"confirmed":false,"hidden":false,"folder_uuids":"bd3c6f2e-3514-47cf-bc42-12db8b9abe45","private_publication":false,"abstract":"Autonomous robotic platforms can be effectively used to perform automatic fruits yield estimation. To this aim, robots need data-driven models that process image streams and count, even approximately, the number of fruits in an orchard. However, training such models following a supervised paradigm is expensive and unpractical. Extending pre-trained models to perform yield estimation for a completely new type of fruit is even more challenging, but interesting since this situation is typical in practice. In this work, we combine a State-of-the-Art weakly-supervised fruit counting model with an unsupervised style transfer method for addressing the task above. In this sense, our proposed approach is quasi-unsupervised. In particular, we use a Cycle-Generative Adversarial Network (C-GAN) to perform unsupervised domain adaptation and train it alongside with a Presence-Absence Classifier (PAC) that discriminates images containing fruits or not. The PAC produces the weak-supervision signal for the counting network, that can then be used on the target orchard directly. Experiments on datasets collected in four different orchards show that the proposed approach is more accurate than the supervised baseline methods.","bibtype":"article","author":"Bellocchio, Enrico and Costante, Gabriele and Cascianelli, Silvia and Fravolini, Mario Luca and Valigi, Paolo","doi":"10.1109/LRA.2020.2966398","journal":"IEEE Robotics and Automation Letters","number":"2","bibtex":"@article{\n title = {Combining Domain Adaptation and Spatial Consistency for Unseen Fruits Counting: A Quasi-Unsupervised Approach},\n type = {article},\n year = {2020},\n keywords = {Agricultural automation,robotics in agriculture and forestry,visual learning},\n pages = {1079-1086},\n volume = {5},\n month = {4},\n publisher = {Institute of Electrical and Electronics Engineers Inc.},\n day = {1},\n id = {4cacbaff-57d3-31f6-a4da-40c99c38d3bb},\n created = {2023-01-06T11:09:15.016Z},\n accessed = {2023-01-06},\n file_attached = {true},\n profile_id = {235249c2-3ed4-314a-b309-b1ea0330f5d9},\n group_id = {5ec9cc91-a5d6-3de5-82f3-3ef3d98a89c1},\n last_modified = {2023-01-06T11:09:22.711Z},\n read = {false},\n starred = {false},\n authored = {false},\n confirmed = {false},\n hidden = {false},\n folder_uuids = {bd3c6f2e-3514-47cf-bc42-12db8b9abe45},\n private_publication = {false},\n abstract = {Autonomous robotic platforms can be effectively used to perform automatic fruits yield estimation. To this aim, robots need data-driven models that process image streams and count, even approximately, the number of fruits in an orchard. However, training such models following a supervised paradigm is expensive and unpractical. Extending pre-trained models to perform yield estimation for a completely new type of fruit is even more challenging, but interesting since this situation is typical in practice. In this work, we combine a State-of-the-Art weakly-supervised fruit counting model with an unsupervised style transfer method for addressing the task above. In this sense, our proposed approach is quasi-unsupervised. In particular, we use a Cycle-Generative Adversarial Network (C-GAN) to perform unsupervised domain adaptation and train it alongside with a Presence-Absence Classifier (PAC) that discriminates images containing fruits or not. The PAC produces the weak-supervision signal for the counting network, that can then be used on the target orchard directly. Experiments on datasets collected in four different orchards show that the proposed approach is more accurate than the supervised baseline methods.},\n bibtype = {article},\n author = {Bellocchio, Enrico and Costante, Gabriele and Cascianelli, Silvia and Fravolini, Mario Luca and Valigi, Paolo},\n doi = {10.1109/LRA.2020.2966398},\n journal = {IEEE Robotics and Automation Letters},\n number = {2}\n}","author_short":["Bellocchio, E.","Costante, G.","Cascianelli, S.","Fravolini, M., L.","Valigi, P."],"urls":{"Paper":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c/file/58969234-e1e3-66db-7ee7-131c1d1093fe/full_text.pdf.pdf"},"biburl":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c","bibbaseid":"bellocchio-costante-cascianelli-fravolini-valigi-combiningdomainadaptationandspatialconsistencyforunseenfruitscountingaquasiunsupervisedapproach-2020","role":"author","keyword":["Agricultural automation","robotics in agriculture and forestry","visual learning"],"metadata":{"authorlinks":{}},"downloads":0},"bibtype":"article","biburl":"https://bibbase.org/service/mendeley/bfbbf840-4c42-3914-a463-19024f50b30c","dataSources":["2252seNhipfTmjEBQ"],"keywords":["agricultural automation","robotics in agriculture and forestry","visual learning"],"search_terms":["combining","domain","adaptation","spatial","consistency","unseen","fruits","counting","quasi","unsupervised","approach","bellocchio","costante","cascianelli","fravolini","valigi"],"title":"Combining Domain Adaptation and Spatial Consistency for Unseen Fruits Counting: A Quasi-Unsupervised Approach","year":2020}