Optimal estimation of broiler movement for commercial tracking
Nowadays, video tracking has taken a considerable part in monitoring systems. It allows identifying and follow every object in the camera field over time. While most of these algorithms are rather well suited to regular movements (following cars, pedestrians), they are often limited in more complex...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-02-01
|
| Series: | Smart Agricultural Technology |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2772375522000788 |
| _version_ | 1818065537827078144 |
|---|---|
| author | Henry Brunet Didier Concordet |
| author_facet | Henry Brunet Didier Concordet |
| author_sort | Henry Brunet |
| collection | DOAJ |
| description | Nowadays, video tracking has taken a considerable part in monitoring systems. It allows identifying and follow every object in the camera field over time. While most of these algorithms are rather well suited to regular movements (following cars, pedestrians), they are often limited in more complex situations (high variations in speed, low detection rate, frequent shape variation). This paper proposes three methods adapted to broilers tracking in commercial environment. Past movements analysis of known broilers enable to estimate their motions and therefore to predict their new position. New unidentified broilers positions are then compared to these predicted positions. Distances between these two sets of positions are then used in the Hungarian Algorithm to assign an ID to new detected broilers regarding their past positions. Our methods differentiate by the way they predict the future positions. Contrary to most methods, they do not seek perfect regularity of movements and can deal with low rate detection. The proposed methods showed better performances than existing one. Tests have been made at 21, 26, and 37 days of age. At 21 days, our best method produces up to 35% fewer errors than a method with no estimation of movement. At 26 days of age, displacement distances can be set to only 68% of the maximum recorded displacement while improving an average of 21% of tracking errors across all methods. |
| first_indexed | 2024-12-10T14:53:28Z |
| format | Article |
| id | doaj.art-5f81c3162581442681843e4a295cc587 |
| institution | Directory Open Access Journal |
| issn | 2772-3755 |
| language | English |
| last_indexed | 2024-12-10T14:53:28Z |
| publishDate | 2023-02-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Smart Agricultural Technology |
| spelling | doaj.art-5f81c3162581442681843e4a295cc5872022-12-22T01:44:22ZengElsevierSmart Agricultural Technology2772-37552023-02-013100113Optimal estimation of broiler movement for commercial trackingHenry Brunet0Didier Concordet1Corresponding author.; Ecole Nationnal Veterinaire de Toulouse, 23 Chemin des Capelles, Toulouse 31300, FranceEcole Nationnal Veterinaire de Toulouse, 23 Chemin des Capelles, Toulouse 31300, FranceNowadays, video tracking has taken a considerable part in monitoring systems. It allows identifying and follow every object in the camera field over time. While most of these algorithms are rather well suited to regular movements (following cars, pedestrians), they are often limited in more complex situations (high variations in speed, low detection rate, frequent shape variation). This paper proposes three methods adapted to broilers tracking in commercial environment. Past movements analysis of known broilers enable to estimate their motions and therefore to predict their new position. New unidentified broilers positions are then compared to these predicted positions. Distances between these two sets of positions are then used in the Hungarian Algorithm to assign an ID to new detected broilers regarding their past positions. Our methods differentiate by the way they predict the future positions. Contrary to most methods, they do not seek perfect regularity of movements and can deal with low rate detection. The proposed methods showed better performances than existing one. Tests have been made at 21, 26, and 37 days of age. At 21 days, our best method produces up to 35% fewer errors than a method with no estimation of movement. At 26 days of age, displacement distances can be set to only 68% of the maximum recorded displacement while improving an average of 21% of tracking errors across all methods.http://www.sciencedirect.com/science/article/pii/S2772375522000788Point trackingBroilers welfare assessmentDisplacement model |
| spellingShingle | Henry Brunet Didier Concordet Optimal estimation of broiler movement for commercial tracking Smart Agricultural Technology Point tracking Broilers welfare assessment Displacement model |
| title | Optimal estimation of broiler movement for commercial tracking |
| title_full | Optimal estimation of broiler movement for commercial tracking |
| title_fullStr | Optimal estimation of broiler movement for commercial tracking |
| title_full_unstemmed | Optimal estimation of broiler movement for commercial tracking |
| title_short | Optimal estimation of broiler movement for commercial tracking |
| title_sort | optimal estimation of broiler movement for commercial tracking |
| topic | Point tracking Broilers welfare assessment Displacement model |
| url | http://www.sciencedirect.com/science/article/pii/S2772375522000788 |
| work_keys_str_mv | AT henrybrunet optimalestimationofbroilermovementforcommercialtracking AT didierconcordet optimalestimationofbroilermovementforcommercialtracking |