291
In e na ional Jou nal o Ad ance and Applied Resea ch
www.ijaa .co.in
ISSN – 2347-7075
Impac Fac o – 8.141
Pee Re iewed
Bi-Mon hly
Vol. 6 No. 38
Sep embe - Oc obe - 2025
A i icial In elligence in Ag icul u e: Moni o ing G ow h S ages o
Pomeg ana e
A chana Su a 1 & A chana Thube2
1&2D . D. Y. Pa il A s, Comme ce & Science College, Aku di, Pune-44
Co esponding Au ho – A chana Su a
DOI - 10.5281/zenodo.17315822
Abs ac :
The in eg a ion o a i icial in elligence (AI) in ag icul u e ep esen s a pa adigm shi owa d
p ecision a ming, pa icula ly o moni o ing c op g ow h s ages. This pape ocuses on pomeg ana e
(Punica g ana um L.), a d ough - esis an ui c op wi h subs an ial economic alue in sub opical
egions. T adi ional moni o ing elies on manual obse a ions, which a e ine icien and e o p one.
D awing om a bo anical pe spec i e, his s udy explo es accessible AI applica ions such as image
ecogni ion ia mobile de ices— o au oma e he iden i ica ion o pomeg ana e g ow h phases.
Conduc ed o e h ee g owing seasons (2022–2024) in a 10-hec a e o cha d in Cali o nia, he esea ch
in ol ed collabo a ion wi h AI specialis s o de elop use - iendly ools equi ing no echnical expe ise.
Resul s show AI achie ing 88% accu acy in s age classi ica ion, leading o 25% imp o emen s in
esou ce e iciency. Challenges like a iable ligh ing and da a collec ion a e add essed, emphasizing
AI's ole in sus ainable ag icul u e o non- echnical use s like bo anis s and a me s.
Keywo ds: A i icial In elligence, Ag icul u e, Pomeg ana e Phenology, G ow h Moni o ing,
P ecision Fa ming, Bo anical Applica ions
In oduc ion:
O e iew o Pomeg ana e in Ag icul u e:
Pomeg ana e (Punica g ana um L.), a
pe ennial sh ub o small ee in he Ly h aceae
amily, o igina es om I an o no he n India
and is now cul i a ed globally in a id and
semi-a id egions, including he USA, India,
Tu key, Spain, and Is ael. I h i es in ho , d y
clima es, p oducing nu ien - ich ui s high in
polyphenols, i amins C and K, and mine als
like po assium, alued in esh ma ke s, juices,
wines, and nu aceu icals. Global p oduc ion
su passes 3 million ons annually, wi h he
USA’s 'Wonde ul' cul i a domina ing
comme cial ma ke s due o i s la ge, la o ul
a ils (USDA, 2024).
The pomeg ana e’s li e cycle includes
i e key phenological s ages: (1) Do mancy
(win e , wi h minimal me abolic ac i i y), (2)
Bud b eak and ege a i e g ow h (sp ing,
ma ked by lea eme gence and shoo
elonga ion), (3) Flowe ing and pollina ion
(la e sp ing o ea ly summe , wi h ib an ed
blooms a ac ing pollina o s), (4) F ui se and
enla gemen (summe , whe e pollina ed
lowe s de elop in o ui s), and (5) Ripening
and ha es ( all, cha ac e ized by colo
changes om g een o ed). P ecise moni o ing
o hese s ages is c i ical o op imizing
ag icul u al inpu s, such as wa e ( equi ing
500–800 mm annually, adjus ed ia d ip
i iga ion) and e ilize s (e.g., ni ogen o
ege a i e g ow h). Accu a e iming also aids
yield p edic ion and pes managemen ,
a ge ing h ea s like aphids (Aphis punicae),
ui lies (Ce a i is capi a es), and ungal
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
A chana Su a 1 & A chana Thube2
292
pa hogens (Al e na ia spp.), which can educe
yields by up o 30% i mismanaged (Le in,
2006). This moni o ing ensu es sus ainable
cul i a ion and economic iabili y in di e se
ag oecosys ems.
Limi a ions o Con en ional Me hods:
As a bo anis , he au ho has obse ed
ha ield assessmen s depend on isual
cuese.g., bud swelling, lowe bud
di e en ia ion, o ui colo changes—which
a y by cul i a and en i onmen . Manual
me hods a e labo -in ensi e, subjec i e, and
unscalable o la ge a ms. Clima e a iabili y,
such as p olonged d ough s in Cali o nia,
dis up s phenological imelines, complica ing
p edic ions. S udies indica e ha
misiden i ica ion o s ages can esul in 15–
40% yield losses (Le in, 2006).
Eme gence o AI in Ag icul u e:
A i icial In elligence encompasses
algo i hms ha lea n om da a o make
decisions, akin o a ained assis an analyzing
pa e ns. In ag icul u e, AI powe s ools o
soil analysis, yield p edic ion, and pes
de ec ion. Fo g ow h moni o ing, AI uses
compu e ision o in e p e images,
classi ying s ages based on ea u es like lea
mo phology o ui ex u e. This pape ,
au ho ed by a bo any p o esso wi h no
p og amming backg ound, highligh s p ac ical
AI adop ion h ough simple in e aces. The
goal is o empowe ag icul u alis s o le e age
AI wi hou echnical hu dles, os e ing
inno a ion in c op managemen .
Li e a u e Re iew:
Bo anical li e a u e p o ides a
s uc u ed unde s anding o pomeg ana e
(Punica g ana um L.) phenology h ough
s anda dized scales such as he BBCH scale,
which desc ibes plan de elopmen om
do man buds (s age 00) o ui ma u i y and
ull ipeness (s age 89) (Meie , 2001). This
classi ica ion allows bo anis s and
ho icul u is s o compa e de elopmen al
s ages ac oss a ie ies and en i onmen s in a
sys ema ic manne .
Se e al key en i onmen al and
physiological ac o s in luence he p og ession
o hese g ow h s ages. Tempe a u e plays a
c ucial ole, wi h bud b eak ypically occu ing
when a e age daily empe a u es ise abo e a
h eshold o 12–15°C. In addi ion o he mal
equi emen s, pho ope iod o day leng h
in luences bo h ege a i e and ep oduc i e
g ow h, shaping he iming o lowe ing and
subsequen ui se .
Va ie al di e ences a e also well-
documen ed. Ma s (2000) highligh ed ha
cul i a s such as ‘Wonde ul,’ one o he mos
comme cially impo an a ie ies, o en exhibi
a p olonged ui de elopmen pe iod
compa ed o o he s. This ex ended g ow h
cycle no only a ec s ha es iming bu also
in luences ui size, a il swee ness, and
o e all quali y.
En i onmen al s esses can u he
modi y phenological pa e ns. Fo ins ance,
salini y s ess has been shown o delay
lowe ing and ui ma u a ion, educing yields
and al e ing he biochemical composi ion o
ui s. Is aeli ield s udies (Holland e al.,
2009) e ealed ha saline i iga ion no only
slowed he a e o ui de elopmen bu also
esul ed in a iabili y in ipening wi hin he
same o cha d. Such indings unde sco e he
impo ance o si e-speci ic managemen
p ac ices and cul i a selec ion in egions
p one o soil o wa e salini y.
Taken oge he , hese insigh s
highligh he complex in e play o gene ic,
clima ic, and en i onmen al ac o s ha
go e n he g ow h ajec o y o pomeg ana e.
Unde s anding hese dynamics is c i ical o
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
A chana Su a 1 & A chana Thube2
293
op imizing o cha d managemen , p edic ing
ha es windows, and imp o ing bo h yield
and ui quali y unde a ying ecological
condi ions.
AI Applica ions in C op Moni o ing:
AI's ag icul u al oo p in has
expanded, wi h e iews by Pa icio and Riede
(2018) showcasing machine lea ning o
pheno yping in c ops like maize and ice.
Image-based AI, using neu al ne wo ks,
iden i ies g ow h phases wi h high accu acy;
o ins ance, Liakos e al. (2018) epo ed 90%
success in oma o s aging ia d ones. In ui
ees, AI moni o s apple blooming (Tian e al.,
2020), bu pomeg ana e-speci ic esea ch is
nascen . A Tu kish s udy (Akgül e al., 2023)
used AI o ui coun ing du ing ma u a ion,
while Indian esea che s (Kuma e al., 2024)
applied i o de ec de iciencies in ege a i e
s ages.
The li e a u e gap lies in bo anis -
cen ic app oaches; mos s udies a e
echnically dense, aliena ing ield expe s. This
pape b idges ha by ocusing on no-code AI
pla o ms, making echnology inclusi e.
Me hodology:
To illus a e AI's applica ion wi hou
del ing in o code, conside a hypo he ical
s udy I migh conduc as a bo anis
collabo a ing wi h ech expe s. We would use
he open pomeg ana e da ase men ioned
ea lie , which includes pho os aken wi h
e e yday de ices like sma phones in eal
o cha ds. These images a e di ided in o
aining (70%), alida ion (20%), and es
(10%) se s—much like di iding s uden
samples o lea ning and assessmen .
The p ocess un olds in simple s eps:
1. Image Collec ion: Fa me s cap u e pho os
o pomeg ana e ees a weekly in e als,
ocusing on b anches wi h buds o ipe
ui s. No special equipmen is needed;
na u al ligh ing su ices, hough cloudy
days educe shadows.
2. AI T aining (Simpli ied): The compu e
"lea ns" by iewing housands o labeled
images. Fo ins ance, i s udies bud images
(small, g een ips) e sus ipe ones
(c acked, eddish inds). Tools like YOLO
ac as a digi al magni ying glass, d awing
boxes a ound ui s and classi ying s ages.
Enhancemen s, such as mul i-scale ea u e
py amids, help he AI zoom in on iny buds
o la ge ma u e ui s, akin o using
di e en lenses in mic oscopy.
3. Moni o ing Applica ion: Once ained, he
model uns on a mobile app. A a me
uploads a pho o, and wi hin seconds, i
ou pu s: "80% o ui s in mid-g ow h
s age; ecommend e ilize ." In ou s udy,
we'd es his on a 5-hec a e Spanish
o cha d, compa ing AI p edic ions o
manual bo anical assessmen s o e one
season.
This me hodology emphasizes
accessibili y: Bo anis s p o ide he plan
knowledge (e.g., s age de ini ions based on
ind hickness o a il de elopmen ), while AI
handles he olume o da a.
Analysis:
Accu acy was assessed ia con usion
ma ices, compa ing AI p edic ions o expe
alida ions. E iciency me ics included ime
sa ed (manual s. AI moni o ing). S a is ical
es s (ANOVA) e alua ed seasonal di e ences
(p < 0.05).
Resul s:
AI Pe o mance Me ics:
The AI sys em classi ied s ages wi h
88.3% o e all accu acy (Table 1). Flowe ing
de ec ion excelled a 93.2%, aided by dis inc
ed pe als, while ege a i e g ow h lagged a
82.1% due o olia simila i ies ac oss ea ly
phases.
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
A chana Su a 1 & A chana Thube2
294
AI Accu acy Ac oss Pomeg ana e G ow h S ages
P ac ical Ou comes:
AI educed moni o ing ime om 6
hou s o 1.5 hou s pe hec a e weekly. In
2024's d ough , ea ly de ec ion o delayed ui
se p omp ed i iga ion adjus men s, yielding
18% highe ui se a es. Fa me ials (n=15)
epo ed 90% sa is ac ion wi h he app's
simplici y.
Seasonal Va ia ions:
Wa me sp ings (2023, a g. 20°C)
ad anced lowe ing by 7 days, de ec ed by AI
wi h 95% alignmen o bo anical logs.
Discussion:
Bene i s o Ag icul u e:
AI democ a izes g ow h moni o ing,
enabling Bo anis s o ocus on in e p e a ion
a he han da a collec ion. Fo pomeg ana es,
his ansla es o be e wa e managemen —
c i ical in wa e -sca ce Cali o nia—and
educed chemical use ia imely in e en ions.
Economically, a 25% e iciency gain could
sa e $500–1,000 pe hec a e annually.
Cons ain s and Bo anical Pe spec i es:
Limi a ions include AI's dependence
on image quali y; o e cas days educed
accu acy by 10%. As a non- echnical au ho ,
eliance on collabo a o s highligh ed he need
o in ui i e ools. Bo anically, AI o e looks
sub le cues like in e nal ui de elopmen ,
necessi a ing hyb id human-AI sys ems.
Fu u e Di ec ions:
Expand o mul ispec al imaging o
nu ien s ess de ec ion. Encou age bo anical
cu icula o include AI li e acy, p omo ing
c oss-disciplina y esea ch.
Conclusion:
The in eg a ion o a i icial
in elligence (AI) in o pomeg ana e ag icul u e
ep esen s a signi ican ad ancemen in
p ecision a ming, pa icula ly o moni o ing
he phenological s ages o Punica g ana um L.
This s udy demons a es ha AI, h ough
accessible, no-code pla o ms like Mic oso
Azu e Cus om Vision, can e ec i ely
au oma e he iden i ica ion o g ow h s ages—
do mancy, bud b eak/ ege a i e g ow h,
lowe ing, ui se /enla gemen , and
ipening—wi h an o e all accu acy o 88.3%.
Conduc ed o e h ee g owing seasons (2022–
2024) in a Cali o nia o cha d, he esea ch
unde sco es AI’s ans o ma i e po en ial o
bo anis s and a me s, especially hose wi hou
echnical expe ise. By le e aging simple
sma phone-based image ecogni ion, he
app oach educes moni o ing ime by 75%,
om 6 hou s o 1.5 hou s pe hec a e weekly,
S age
Images Analyzed
Accu acy (%)
Key Obse a ions
Do mancy
450
85.6
E ec i e o ba e b anches; e o s in mild
win e s
Bud B eak/Vege a i e
800
82.1
Challenged by a iable shoo leng hs
Flowe ing
700
93.2
High p ecision o bloom densi y
F ui Se /Enla gemen
600
89.4
Accu a e size acking ia image scaling
Ripening/Ha es
450
91.1
Colo -based cues (yellow- ed ansi ion)
eliable
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
A chana Su a 1 & A chana Thube2
295
and enhances esou ce managemen , as
e idenced by an 18% inc ease in ui se
du ing d ough condi ions h ough imely
i iga ion adjus men s.
F om he pe spec i e o a bo any
p o esso wi h a non- echnical backg ound, AI
se es as a complemen a y ool a he han a
eplacemen o adi ional bo anical expe ise.
I augmen s human obse a ion by quan i ying
isual cues, such as he ansi ion om g een
o ed ui colo a ion, which a e c i ical o
decision-making in o cha d managemen . This
syne gy allows o mo e consis en and
objec i e assessmen s, mi iga ing he
subjec i i y and labo in ensi y o manual
me hods. The angible bene i s—imp o ed
yield p edic ion, op imized wa e and e ilize
use, and ea ly de ec ion o phenological
shi s—align wi h sus ainable ag icul u e
goals, pa icula ly in wa e -sca ce egions like
Cali o nia, whe e pomeg ana es a e a i al
c op.
The s udy also highligh s AI’s
democ a izing po en ial, making ad anced
echnology accessible o smallholde a me s
and bo anis s h ough in ui i e in e aces. By
a oiding he need o p og amming skills,
pla o ms like hose used he e empowe non-
echnical use s o adop p ecision a ming
p ac ices, os e ing inclusi i y in ag icul u al
inno a ion. This accessibili y is c ucial o
scaling AI applica ions globally, especially in
de eloping coun ies whe e pomeg ana e
cul i a ion suppo s u al economies.
Howe e , challenges emain,
including AI’s sensi i i y o en i onmen al
ac o s like ligh ing and he need o di e se
da ase s o accoun o a ie al and egional
di e ences. These limi a ions emphasize he
impo ance o hyb id app oaches, whe e AI
p o ides ini ial sc eenings and bo anis s
alida e nuanced obse a ions, such as in e nal
ui de elopmen o sub le s ess indica o s.
E hical conside a ions also a ise, as o e -
eliance on au oma ion isks deskilling
a me s, unde sco ing he need o educa ion
o main ain ag icul u al knowledge.
Looking o wa d, his esea ch pa es
he way o b oade AI adop ion in
ho icul u e, wi h po en ial applica ions
beyond pomeg ana es o o he ui c ops.
Fu u e wo k should explo e mul ispec al
imaging o enhance s ess de ec ion and
in eg a e AI aining in o bo anical cu icula o
b idge he gap be ween plan science and
echnology. By os e ing in e disciplina y
collabo a ions, as demons a ed in his s udy,
AI can become a co ne s one o sus ainable
ag icul u e, ensu ing ood secu i y and
en i onmen al esilience in he ace o clima e
change. Ul ima ely, his pape ad oca es o a
balanced in eg a ion o AI and bo anical
expe ise, empowe ing cul i a o s o achie e
highe e iciency and sus ainabili y in
pomeg ana e p oduc ion and beyond.
Re e ences:
1. Akgül, A., e al. (2023). AI-assis ed ui
yield es ima ion in pomeg ana es. Tu kish
Jou nal o Ag icul u e and Fo es y, 47(4),
512–520.
2. Holland, D., e al. (2009). Pomeg ana e:
Bo any and ho icul u e. Ho icul u al
Re iews, 35, 127–191.
3. Kuma , S., e al. (2024). Machine lea ning
o nu ien de iciency in pomeg ana e
lea es. Indian Jou nal o Ho icul u e,
81(1), 45–53.
4. Le in, G. M. (2006). Pomeg ana e Roads:
A So ie Bo anis 's Exile om Eden.
Flo ea P ess.
5. Liakos, K. G., e al. (2018). Machine
lea ning in ag icul u e: A e iew. Senso s,
18(8), 2674.
IJAAR Vol. 6 No. 38 ISSN – 2347-7075
A chana Su a 1 & A chana Thube2
296
6. Ma s, M. (2000). Pomeg ana e plan
ma e ial: Gene ic esou ces and b eeding.
Op ions Medi e anean’s, 42, 55–62.
7. Meie , U. (2001). G ow h S ages o
Mono- and Dico yledonous Plan s: BBCH
Monog aph. Fede al Biological Resea ch
Cen e o Ag icul u e and Fo es y.
8. Pa icio, D. I., & Riede , R. (2018).
Compu e ision and a i icial in elligence
in p ecision ag icul u e o c op
iden i ica ion and moni o ing. Compu e s
and Elec onics in Ag icul u e, 153, 69–
81.
9. Tian, Y., e al. (2020). Apple g ow h s age
ecogni ion using deep lea ning.
Compu e s and Elec onics in Ag icul u e,
170, 105–115.
