Learning an optical filter for green pepper automatic picking in agriculture

2021 ◽  
Vol 191 ◽  
pp. 106521
Author(s):  
Xinzhi Liu ◽  
Jun Yu ◽  
Toru Kurihara ◽  
Ke Li ◽  
Zhao Niu ◽  
...  
Sensors ◽  
2021 ◽  
Vol 21 (19) ◽  
pp. 6437
Author(s):  
Jun Yu ◽  
Toru Kurihara ◽  
Shu Zhan

There is a growing demand for developing image sensor systems to aid fruit and vegetable harvesting, and crop growth prediction in precision agriculture. In this paper, we present an end-to-end optimization approach for the simultaneous design of optical filters and green pepper segmentation neural networks. Our optimization method modeled the optical filter as one learnable neural network layer and attached it to the subsequent camera spectral response (CSR) layer and segmentation neural network for green pepper segmentation. We used not only the standard red–green–blue output from the CSR layer but also the color-ratio maps as additional cues in the visible wavelength and to augment the feature maps as the input for segmentation. We evaluated how well our proposed color-ratio maps enhanced optical filter design methods in our collected dataset. We find that our proposed method can yield a better performance than both an optical filter RGB system without color-ratio maps and a raw RGB camera (without an optical filter) system. The proposed learning-based framework can potentially build better image sensor systems for green pepper segmentation.


2002 ◽  
Vol 722 ◽  
Author(s):  
T. S. Sriram ◽  
B. Strauss ◽  
S. Pappas ◽  
A. Baliga ◽  
A. Jean ◽  
...  

AbstractThis paper describes the results of extensive performance and reliability characterization of a silicon-based surface micro-machined tunable optical filter. The device comprises a high-finesse Fabry-Perot etalon with one flat and one curved dielectric mirror. The curved mirror is mounted on an electrostatically actuated silicon nitride membrane tethered to the substrate using silicon nitride posts. A voltage applied to the membrane allows the device to be tuned by adjusting the length of the cavity. The device is coupled optically to an input and an output single mode fiber inside a hermetic package. Extensive performance characterization (over operating temperature range) was performed on the packaged device. Parameters characterized included tuning characteristics, insertion loss, filter line-width and side mode suppression ratio. Reliability testing was performed by subjecting the MEMS structure to a very large number of actuations at an elevated temperature both inside the package and on a test board. The MEMS structure was found to be extremely robust, running trillions of actuations without failures. Package level reliability testing conforming to Telcordia standards indicated that key device parameters including insertion loss, filter line-width and tuning characteristics did not change measurably over the duration of the test.


Author(s):  
Agustina Onyebuchi Ijeomah ◽  
Rebecca Ngoholve Vesuwe ◽  
Bitrus Pam

Vegetables growing in mining areas have become a serious food safety concern because of the high levels of heavy metals always associated with mining. In this study, water used for irrigation, soil, cabbage, green pepper and green beans grown in tin mine areas of Heipang District, Barkin-Ladi LGA of Plateau State were analyzed for lead, cadmium and zinc, using Atomic Absorption Spectrophotometer (AAS). The concentrations of the heavy metals in water, soil, vegetables were all in the order Pb, >> Cd > Zn. In the vegetables, the order was: Pb → cabbage > green beans > green pepper; Cd → green beans > cabbage > green pepper; Zn → cabbage > green pepper = green beans. The transfer factors for all the metals (heavy metal in plant / heavy metal in soil) ranged from 0.95 to 1.48. There were high levels of Pb and Cd in all the vegetables, which may be attributed to the metals in the water used for irrigation. Whilst the concentration of Zn in all the samples were lower than recommended limits, the levels of Pb and Cd in the water, soil and vegetables were higher than the WHO/FEPA standard recommended limits reported for vegetables. The Cd concentrations of the vegetables also exceeded the tolerance thresholds for animals and human beings and therefore consumption of vegetable from the area would endanger the health of the population.


HortScience ◽  
1998 ◽  
Vol 33 (3) ◽  
pp. 550d-550
Author(s):  
Eric H. Simonne ◽  
John T. Owen

The retail value of yellow and red bell peppers is usually three to five times higher than that of the green ones. However, colored bell pepper production in Alabama is presently limited because most growers do not wait the additional 3 to 6 days needed for marketable green pepper to develop color. Hence, drip-irrigated yellow `Admiral' and `Goldcoast' and red `Bell Star' and `Capsitrano' bell peppers were grown in single row and bare-ground, and harvested as needed between July and October 1997 at the 0/3 (green), 1/3 or 2/3 colored stages. The interaction variety × picking method was not signficant (P > 0.50). Early (9,136 kg/ha) and total (32,363 kg/ha) yields of green (0/3) peppers were significantly (P < 0.05) higher than those of the 1/3 and 2/3 colored ones (5,166 and 27,235 kg/ha, respectively). Higher green yields were mainly due to increased numbers of marketable fruits rather than increased fruit size. The longer the pods stayed on the plants, the more likely was sunscald to occur. Retail values (/ha) for the early fancy grade were $10,800 and $20,500 for the green and colored peppers, respectively (using $2 and $6/kg, respectively). These results suggest that the present higher retail value of the colored bell peppers off-sets the lower expected yields.


Author(s):  
Saidajan Attiq Abdiani ◽  
Kifayatullah Kakar ◽  
Gulbuddin Gulab ◽  
Shafiqullah Aryan

Machines ◽  
2021 ◽  
Vol 9 (3) ◽  
pp. 66
Author(s):  
Tianci Chen ◽  
Rihong Zhang ◽  
Lixue Zhu ◽  
Shiang Zhang ◽  
Xiaomin Li

In an orchard environment with a complex background and changing light conditions, the banana stalk, fruit, branches, and leaves are very similar in color. The fast and accurate detection and segmentation of a banana stalk are crucial to realize the automatic picking using a banana picking robot. In this paper, a banana stalk segmentation method based on a lightweight multi-feature fusion deep neural network (MFN) is proposed. The proposed network is mainly composed of encoding and decoding networks, in which the sandglass bottleneck design is adopted to alleviate the information a loss in high dimension. In the decoding network, a different sized dilated convolution kernel is used for convolution operation to make the extracted banana stalk features denser. The proposed network is verified by experiments. In the experiments, the detection precision, segmentation accuracy, number of parameters, operation efficiency, and average execution time are used as evaluation metrics, and the proposed network is compared with Resnet_Segnet, Mobilenet_Segnet, and a few other networks. The experimental results show that compared to other networks, the number of network parameters of the proposed network is significantly reduced, the running frame rate is improved, and the average execution time is shortened.


2021 ◽  
Vol 53 (5) ◽  
Author(s):  
Asmaa Ibrahim ◽  
Josep Prat ◽  
Tawfik Ismail

Sign in / Sign up

Export Citation Format

Share Document