Identification Method of Planktonic Algae Community Based on Multi-Task Convolutional Neural Network

Zhao Cheng; Nanjing Zhao; Gaofang Yin; Xiaoling Zhang; Xiang Wang

doi:10.3788/AOS202242.0530002

Acta Optica Sinica, Volume. 42, Issue 5, 0530002(2022)

Identification Method of Planktonic Algae Community Based on Multi-Task Convolutional Neural Network

Zhao Cheng^1,2,3, Nanjing Zhao^1,3、*, Gaofang Yin^1,3, Xiaoling Zhang⁴, and Xiang Wang^1,2,3

¹Key Laboratory of Environmental Optics and Technology, Anhui Institute of Optics and Fine Mechanics, Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei, Anhui 230031, China

²University of Science and Technology of China, Hefei, Anhui 230026, China

³Key Laboratory of Environmental Optical Monitoring Technology of Anhui Province, Hefei, Anhui 230031, China

⁴Anhui University, Hefei, Anhui 230601, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(12)

Fig. 1. Discrete three-dimensional fluorescence spectra of different algae. (a) Microcystis aeruginosa;(b) Scenedesmus obliquus;(c) Nitzschia sp.; (d) Peridinium umbonatum var.inaequale; (e) Cryptomonas obovata.; (f) Microcystis aeruginosa+Scenedesmus obliquus; (g) Scenedesmus obliquus+Peridinium umbonatum var.inaequale; (h) Nitzschia sp.+Peridinium umbonatum var.inaequale+Cryptomonas obovata.; (i) Microcystis aeruginosa+Scenedesmus obliquus+Nitzschia sp.+Peridinium umbonatum var.inaequale; (j) Microcystis aeruginosa+Scenedesmus obliquus+Nitzschia sp.+Peridinium umbonatum var.inaequale+Cryptomonas obovata.

Download full size

Fig. 2. Structural diagram of TextCNN

Download full size

Fig. 3. R_loss of training results of algal classification

Download full size

Fig. 4. R_loss of regression training by TextCNN model

Download full size

Fig. 5. R_loss of training by PlainCNN-MT model

Download full size

Table 1. Experimental algae
View table
Table 1. Experimental algae
Phylum Alga
Cyanophyta Microcystis aeruginosa
Chlorophyta Scenedesmus obliquus
Bacillariophyta Nitzschia sp.
Pyrroptata Peridinium umbonatum var.inaequale
Cryptophyta Cryptomonas obovata.

Table 2. Concentrations of algal samples in verification setunit:μg·L^-1

View table

Table 2. Concentrations of algal samples in verification setunit:μg·L^-1

No.	Microcystis aeruginosa	Scenedesmus obliquus	Nitzschia sp.	Peridinium umbonatumvar.inaequale	Cryptomonasobovata.
1	25.98	0	0	0	0
2	0	32.16	0	0	0
3	46.93	0	0	0	10.68
4	0	27.36	10.42	0	0
5	0	0	0	8.77	55.26
6	0	50.91	0	6.58	0
7	17.29	0	0	40.25	0
8	34.17	99.05	0	28.3	0
9	0	46.93	0	54.06	47.92
10	47.52	0	74.18	34.69	0
11	38.25	0	63.73	0	33.97
12	34.15	62.43	0	0	22.66
13	26.19	71.32	87.64	69.27	0
14	0	96.58	84.06	59.11	38.64
15	42.94	126.13	83.85	60.71	0
16	35.64	0	89.37	50.74	46.37
17	64.78	96.53	43.08	0	40.27
18	48.41	76.53	53.19	79.64	21.05
19	33.95	80.26	74.83	46.19	65.72
20	36.05	97.24	46.52	20.47	0

Table 3. Results of classification models
View table
Table 3. Results of classification models
Sample Sample number Accuracy of TextCNN /% Accuracy of PlainCNN /%
Pure 2 100.00 100.00
Two mixed 5 100.00 100.00
Three mixed 5 100.00 100.00
Four mixed 6 66.66 83.33
Five mixed 2 50.00 50.00
Average 85.00 90.00

Table 4. Regression analysis results by TextCNN model

View table

Table 4. Regression analysis results by TextCNN model

No.	Concentration of Microcystisaeruginosa /(μg·L^-1)	Concentration of Scenedesmusobliquus /(μg·L^-1)	Concentration of Nitzschia sp. /(μg·L^-1)	Concentration of Peridinium umbonatumvar.inaequale /(μg·L^-1)	Concentration of Cryptomonasobovata. /(μg·L^-1)	R_MSE
1	24.14	0	0	0	0	0.008
2	0	34.24	0	0	0
3	41.62	0	0	0	11.83
4	0	25.97	12.53	0	0
5	0	0	0	9.57	52.49	0.027
6	0	47.18	0	6.14	0
7	16.29	0	0	37.32	0
8	37.26	95.35	0	26.53	0
9	0	42.81	0	59.47	49.27
10	51.24	0	78.26	33.19	0	0.059
11	40.72	0	66.41	0	30.67
12	32.38	60.34	0	0	19.55
13	23.14	72.42	85.49	72.96	0
14	0	96.05	82.31	55.83	41.28
15	45.03	122.64	86.27	57.46	0	0.084
16	39.14	0	93.14	48.15	41.96
17	59.48	91.57	40.74	0	44.35
18	44.17	73.64	56.39	82.63	18.27
19	30.76	83.16	71.25	42.17	69.43	0.130
20	39.59	101.76	42.61	23.17	37.24
Average						0.062

Table 5. Regression analysis results by PlainCNN model

View table

Table 5. Regression analysis results by PlainCNN model

No.	Concentration of Microcystisaeruginosa /(μg·L^-1)	Concentration of Scenedesmusobliquus /(μg·L^-1)	Concentration of Nitzschia sp. /(μg·L^-1)	Concentration of Peridinium umbonatumvar.inaequale /(μg·L^-1)	Concentration of Cryptomonasobovata. /(μg·L^-1)	R_MSE
1	23.69	0	0	0	0	0.007
2	0	30.58	0	0	0
3	50.37	0	0	0	9.65
4	0	29.83	9.37	0	0
5	0	0	0	9.35	57.28	0.017
6	0	53.05	0	6.74	0
7	18.63	0	0	43.69	0
8	36.28	94.28	0	30.16	0
9	0	48.39	0	50.84	45.27
10	43.16	0	72.51	37.57	0	0.049
11	39.14	0	59.14	0	31.05
12	32.07	64.12	0	0	24.71
13	30.19	69.07	84.38	67.18	0
14	0	93.05	81.54	63.74	40.32
15	45.23	130.26	80.92	61.25	0	0.077
16	33.94	0	92.65	54.29	50.82
17	59.82	98.35	43.83	0	37.48
18	44.87	73.84	56.87	83.46	25.19
19	37.16	77.31	78.49	43.81	61.15	0.109
20	32.86	101.27	49.15	23.07	32.58
Average						0.052

Table 6. Results of classification training by PlainCNN-MT model
View table
Table 6. Results of classification training by PlainCNN-MT model
Sample Sample number Accuracy /%
Pure 2 100
Two mixed 5 100
Three mixed 5 100
Four mixed 6 100
Five mixed 2 50
Average 95

Table 7. Analysis results by PlainCNN-MT model

View table

Table 7. Analysis results by PlainCNN-MT model

No.	Concentration of Microcystisaeruginosa /(μg·L^-1)	Concentration of Scenedesmusobliquus /(μg·L^-1)	Concentration of Nitzschia sp. /(μg·L^-1)	Concentration of Peridinium umbonatumvar.inaequale /(μg·L^-1)	Concentration of Cryptomonasobovata. /(μg·L^-1)	R_MSE
1	27.51	0	0	0	0	0.004
2	0	33.57	0	0	0
3	49.15	0	0	0	9.84
4	0	29.16	9.28	0	0
5	0	0	0	8.41	54.78	0.009
6	0	52.67	0	6.37	0
7	15.82	0	0	42.69	0
8	33.64	102.6	0	26.17	0
9	0	43.59	0	52.39	51.39
10	51.49	0	76.52	32.15	0	0.036
11	39.57	0	60.67	0	35.02
12	35.89	64.18	0	0	21.64
13	28.9	73.22	85.49	72.48	0
14	0	99.47	79.36	62.53	36.87
15	40.63	121.53	85.11	59.47	0	0.064
16	37.39	0	92.01	53.9	49.53
17	60.29	98.13	42.58	0	42.71
18	45.31	73.85	50.37	82.78	24.69
19	36.86	77.25	77.41	48.94	63.87	0.080
20	33.25	94.28	43.28	17.94	34.17
Average						0.039

Tools

Get Citation

Copy Citation Text

Zhao Cheng, Nanjing Zhao, Gaofang Yin, Xiaoling Zhang, Xiang Wang. Identification Method of Planktonic Algae Community Based on Multi-Task Convolutional Neural Network[J]. Acta Optica Sinica, 2022, 42(5): 0530002

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites