Problem Statement: Smart AI-Based Agricultural

Assistance for End-to-End Crop Support

Motivation:

We understand that farming is not just a profession; it's a way of life, deeply rooted in tradition
and essential for sustaining communities. By leveraging technology and innovation, we strive to
empower farmers with the tools they need to overcome obstacles and thrive in an ever-changing
agricultural landscape.

Objective:

To create and implement an intelligent farm support system that aids farmers at every stage of
the crop growing process, right from soil quality testing, to real-time monitoring of crops and
soil health, until end-of-cycle reporting and learning. The system will employ IoT sensors,
computer vision, and AI-driven decision-making to facilitate maximum yield, early problem
detection, and improvement in farming methods over the long term.

Phase-wise Breakdown
Phase 1: Soil Quality Assessment and Crop Feasibility

Use IoT sensors to measure essential soil parameters such as pH, nitrogen content, phosphorus,
potassium, and moisture, such that the farmer can have remote access to soil quality data while
being at home.

Based on the analysis, check whether the soil is suitable for the farmer's desired crop.

If not feasible, recommend appropriate soil improvement techniques (e.g., composting, crop
rotation, bio-fertilizers).

Phase 2: Continuous Crop and Soil Health Monitoring

Use camera-based systems to visually monitor plant health and detect symptoms of diseases
using AI-driven image classification.

Reassess soil nutrient levels periodically using sensors.

If disease or nutrient deficiency is detected, suggest appropriate treatment (e.g., fertilizers, pest
control, organic remedies) using a trained AI model.

Maintain health logs and updates for each plant batch.

Phase 3: Intelligent Reporting and Learning

●​ At the end of the crop cycle, generate a comprehensive report that includes:​

-​ Soil quality changes

-​ Disease occurrences and treatment responses
-​ Crop yield summary
-​ Recommendations and best practices learned during the cycle​

●​ Help farmers use the report as a learning tool for future decision-making and yield
improvement.​

Expected Impact

-​ Minimize crop failure risks due to soil incompatibility or late disease detection.

-​ Enhance soil health and sustainability through data-driven guidance.

-​ Empower farmers with AI tools to make informed decisions and adopt modern
agricultural practices.​

Methodology ( Work /Research done in the given thrust area )

Phase 1:

To monitor soil health remotely, IoT sensors are deployed in the field to measure key parameters
such as pH, nitrogen, phosphorus, potassium, moisture, and temperature

Based on the collected data, the system checks if the soil conditions meet the specific
requirements of the crop the farmer intends to grow. This is done by comparing live sensor
readings with a crop-specific database that includes ideal ranges for pH, moisture, and nutrient
levels. If the soil meets the criteria, the crop is marked as feasible.

If the analysis shows that the soil is not suitable for the chosen crop, the system recommends
practical soil improvement methods. These include using compost or bio-fertilizers, adjusting pH
with lime or sulfur, or adopting crop rotation techniques to naturally restore soil fertility and
balance.

LINK to access papers on the above topic :

[Link] [Link] [Link] ​

Phase 2:

1. Visual Crop Health Monitoring Using Cameras and AI:

To continuously monitor plant health, cameras should be installed in the field to capture images
of crops at regular intervals or on demand. These images are analyzed using AI-driven image
classification models trained on a custom plant disease dataset. The system is capable of
detecting early symptoms such as leaf spots, leaf curl, wilting, and pest infestations. When an
issue is detected, the system automatically flags the unhealthy plant and sends alerts to the
farmer's mobile device. The alert includes the disease name, a confidence score, and images
highlighting the affected areas. This setup relies on a tech stack including cameras,
TensorFlow-based CNN models, and cloud storage for data backup and analysis.

2. Real-Time Soil Health Monitoring:

IoT soil sensors are integrated into the system to continuously monitor key parameters like pH,
nitrogen, moisture, and more. These sensors are connected to a microcontroller, such as an
Arduino, equipped with wireless capabilities. The data is transmitted in real-time to a cloud
platform, making it accessible to farmers through a web dashboard or mobile application. This
allows farmers to monitor their soil health remotely from home and make informed decisions
based on live data.

3. AI-Based Treatment Recommendations:

When the system detects a disease or a nutrient deficiency, a trained AI model provides targeted
treatment suggestions. These include chemical fertilizers, pest control methods like insecticides
or organic options such as neem oil, and soil correction techniques such as composting or the use
of gypsum. The AI matches symptoms and sensor readings to known treatment datasets to
generate the best solution. For example, if yellowing leaves and low nitrogen levels are detected,
the system may recommend using urea or compost tea. Each recommendation includes dosage,
application timing, and safety precautions to ensure effective and sustainable treatment.

4. Health Logging and Batch History Management:

To support long-term analysis and future planning, the system maintains a digital log for each
crop batch or specific field section. This log records crucial information such as sensor readings,
disease occurrences, treatments applied, and the current stage of the crop. These records help
farmers track patterns, assess the success of interventions, and build a knowledge base for better
yields in future cycles.

LINK to access papers on soil quality:

 [Link] [Link]

LINK to access papers on Plant Leaf Disease Detection:

Technical papers

Phase 3:

At the end of each crop cycle, the system automatically generates a comprehensive report using
data collected throughout the season. This report gives the farmer clear insights into soil quality
changes by comparing parameters like pH, nitrogen, phosphorus, potassium, and moisture from
the start to the end of the cycle. It documents all detected diseases, the dates they occurred,
treatments applied, and the effectiveness of those treatments, such as recovery time or percentage
of plants saved. Additionally, the report includes a crop yield summary with total production,
estimated income, and a comparison with expected results. Based on AI analysis, personalized
recommendations are also provided to improve future crop cycles, such as adjusting nutrient
levels or reusing effective organic treatments like neem oil.

To enhance learning, the system presents insights using easy-to-understand graphs and
summaries. These include trend graphs showing how soil or nutrient levels changed over time
and performance highlights that identify what worked well and what needs improvement. All
reports are auto-saved to a cloud or app-based system, allowing farmers to revisit past seasons
and track their progress over time, ultimately helping them make smarter, data-driven farming
decisions.

Work Done
We have collected and curated our own, a real-world dataset comprising images of chili leaves
categorized into 5 distinct classes. To evaluate the most suitable model for accurate and efficient
disease classification, we conducted a comparative study using several deep learning
architectures: VGG19, AlexNet, MobileNetV3, and DenseNet.

Each model was trained and validated on the dataset, and their performance was assessed using
key evaluation metrics such as accuracy, precision, recall, F1-score, and inference time. Through
this analysis, we were able to identify the model that not only delivered the highest classification
accuracy but also met practical requirements in terms of computational efficiency and suitability
for deployment in a real-time agricultural monitoring system.
The results of this study have been documented in a comprehensive report, which includes model
comparisons, visualizations of training progress, confusion matrices, and an explanation of why
the selected model is best suited for our application. This analysis supports the development of a
robust AI-powered crop health monitoring tool that can accurately detect and classify common
diseases in chili plants, ultimately assisting farmers in early diagnosis and treatment.

Conclusion

In conclusion, after conducting thorough research and analyzing the current demands and gaps in
the field, as well as studying various published papers and existing work over these three phases,
our aim is clear: to propose a comprehensive, one-stop solution that addresses all the identified
gaps. Our goal is to create a system that caters exclusively to the needs and challenges faced by
farmers who wish to cultivate their native crops on their land.

The proposed intelligent agricultural assistant system offers a holistic, technology-driven

approach to modern farming by integrating IoT, AI, and machine learning. Through its three
structured phases—soil assessment, continuous monitoring, and end-of-cycle reporting—it
empowers farmers with timely insights and actionable recommendations. By remotely
monitoring soil parameters and crop health, providing AI-guided treatments, and maintaining a
detailed historical record, the system not only enhances crop yield and soil sustainability but also
equips farmers with valuable knowledge for future planning. Ultimately, this solution promotes
precision agriculture, reduces resource wastage, and supports farmers in making informed,
data-driven decisions for improved productivity and long-term agricultural success.
💰 Estimated Cost Breakdown (in ₹)
Component Qty Unit Total Price Notes
Price (₹)

ESP32 Development Board 1 ₹500 ₹500 Wi-Fi + Bluetooth

Capacitive Soil Moisture 1 ₹120 ₹120 Accurate, long-lasting

Sensor

Soil pH Sensor Kit 1 ₹1200 ₹1200 Analog or UART

NPK Sensor (Optional) 1 ₹3500 ₹3500 Skip for MVP if

needed

DHT22 Temp + Humidity 1 ₹150 ₹150 Air data only

Jumper Wires + Breadboard 1 set ₹150 ₹150 Wiring essentials

Plastic Enclosure (Waterproof) 1 ₹300 ₹300 Outdoor protection

Li-ion Battery + 18650 Holder 1 ₹350 ₹350 Mobile power

Solar Panel + TP4056 Charger 1 ₹500 ₹500 Optional for off-grid

Firebase/Blynk Cloud (Free 1 ₹0 ₹0 Basic tier sufficient

Tier) acct

🆕 Additional Recommended Sensors

Sensor Qty Unit Price Total Notes
(₹) Price

Soil EC Sensor 1 ₹1200–₹15 ₹1300 Analog signal

00

Soil Temperature (DS18B20) 1 ₹200 ₹200 Waterproof, digital

Light Sensor (BH1750) 1 ₹200 ₹200 I2C digital

Rain Sensor (YL-83) 1 ₹120 ₹120 Analog or digital

BME280 (Temp + Humidity + 1 ₹300 ₹300 I2C digital

Pressure)
RTC (DS3231) 1 ₹150 ₹150 Timekeeping for logging

GPS Module (NEO-6M) 1 ₹500–₹600 ₹550 Optional; for mapping &

geotagging

⚙️ Optional Expansion Modules

Module Qty Unit Price Total Use
(₹) Price

External ADC (ADS1115) 1 ₹150–₹250 ₹200 More accurate analog

readings

I2C Multiplexer 1 ₹250 ₹250 If multiple I2C devices

(TCA9548A) conflict

MicroSD Card Module + 1 ₹150–₹250 ₹200 Offline data storage

Card

📦 Final System Cost Estimate

Category Total Price (₹)

Basic System (Original) ₹6820 (with NPK sensor)

New Sensors (7 items) ₹3020

Expansion Modules (3 items) ₹650

✅ Total Estimated Cost (Full Setup) ₹10,490

💡 MVP Version (without NPK, GPS, or solar panel): ₹6,440
●​ Exclude: NPK sensor (₹3500), GPS (₹550), Solar (₹500)​

🔚 Summary
 ●​ ✅ Full system with advanced sensors & offline features = ~₹10,500​
●​ 🚀 Lean MVP version (excluding costly sensors/modules) = ~₹6,500​

●​ 🔄 Easily upgradable later with modular design (I2C multiplexer, ADC, GPS)​

Optimal Soil Requirements and Recommendations for Major Indian Crops

Rice
●​ Soil pH: 5.5–7.0 (optimally slightly acidic to neutral)[Link]. Paddy is
tolerant of a wide range but grows best near pH 6.5.​

●​ N–P–K: ~50:12:12 kg/[Link] (≈120:30:30 kg/ha) applied as urea (N),

SSP/DAP (P), and MOP (K). Basal application supplies one-third of N and all P and
[Link].​

●​ Soil moisture: Require flooded conditions initially. Keep standing water ≈5–10 cm deep
for the first 10–14 days after transplanting, then allow drainage before
[Link]. Maintain moist (puddled) soil until grain fill.​

●​ Soil type: Heavy alluvial/clay loams with high water [Link].

Avoid coarse sands. Clay-loam paddy soils are ideal for moisture-holding.​

●​ Special needs: Rice is often Zn-deficient; apply ZnSO₄·7H₂O ~25 kg/acre at puddling to
avoid Zn [Link]. Iron deficiency (yellowing) from dry conditions is
corrected with foliar FeSO₄. Organic matter (green manure/FYM) improves yield and N
availability.​

●​ Adjustment tips: If soil is acidic (pH <5.5), apply lime

(calcite/dolomite)[Link]. If pH is high (>7.5), elemental sulfur or
gypsum can help. For low N, apply urea (e.g. 100 kg/acre yields ~50 N) or incorporate
legumes/FYM; for low P, add SSP/DAP; for low K, use MOP.​

Wheat
●​ Soil pH: 6.0–7.0 (neutral to mildly acidic)[Link]. Very acidic or
strongly alkaline soils reduce nutrient uptake.​

●​ N–P–K: ≈50:25:12 kg/[Link] (≈125:60:30 kg/ha). Typically applied as

110 kg urea (N), 55 kg DAP/SSP (P), 155 kg MOP (K) per [Link].​

●​ Soil moisture: Requires moderate moisture. Irrigate at crown-root initiation (≈20–25

DAS), tillering (40–45 DAS), booting (60–65 DAS) and flowering (80–85
DAS)[Link]. These stages are most sensitive to drought. Avoid waterlogging.​
 ●​ Soil type: Well-drained clay-loam or loam soils with good structure and moderate
water-holding capacity are [Link]. Heavy black soils work if drainage is
good; very light sands are poor.​

●​ Special needs: Zinc is important (apply ~20 kg ZnSO₄/acre as basal)[Link].

Sulfur may be needed on S-deficient soils (add gypsum). FYM or compost at 5–10 t/ha
boosts fertility and N.​

●​ Adjustment tips: For low pH, lime; for high pH, gypsum/S. If N is low, topdress urea in
splits; if P low, SSP/DAP at sowing; if K low, add MOP. Ensure seed inoculants or
organics for balanced nutrition.​

Maize
●​ Soil pH: 6.0–7.5 (neutral to slightly alkaline)[Link]. Maize grows well
in loamy soils of moderate pH.​

●​ N–P–K: ~35–50:12–24:8–12 kg/[Link] (≈90–125:30–60:20–30 kg/ha).

Typical practice: apply 75–110 kg urea, 75 kg SSP (or 27–55 kg DAP), and 15–20 kg
MOP per [Link].​

●​ Soil moisture: Moderate; avoid waterlogging. Give good moisture at planting, knee
height (≈35–40 DAS), and flowering (silking). Irregular drought at these stages cuts
[Link]. Typically irrigate every 7–10 days in hot seasons.​

●​ Soil type: Deep, well-drained loam or silty-loam soils. Resistant to poor drainage or
stagnant water.​

●​ Special needs: Zinc and magnesium deficiencies are common in Indian

[Link]. Basal ZnSO₄·7H₂O (~8 kg/acre) prevents Zn
[Link]. Mg (apply kieserite/Epsom salt if needed) and micronutrient
blends (e.g. 25 kg/acre) address Fe and other trace [Link].​

●​ Adjustment tips: If N is low, side-dress urea (split; some at sowing, rest at mid-growth).
If P or K low, apply SSP/DAP or MOP at planting. Maintain organic matter (FYM,
compost) for better nutrient retention.​
Cotton
●​ Soil pH: ~6.0–7.5 (neutral to slightly alkaline)[Link]. Prefers neutral
soils; avoid very acid soils.​

●​ N–P–K: ~49:24:40 kg/[Link] (≈120:60:100 kg/ha). This is

commonly given as basal N (14 kg/acre) and P (52 kg DAP), split-N (35 kg/acre
mid-season), and K (20 kg MOP basal + 20 kg split)[Link].​

●​ Soil moisture: Moderate. Cotton is drought-tolerant but benefits from steady moisture
during flowering and boll-setting. Do not waterlog. Irrigate 4–6 times in deep soils, 6–8
in sandy soils, timing as for rice fallows (March–July planting)[Link].​

●​ Soil type: Deep, well-drained alluvial or black (vertisol) [Link].

Avoid poorly drained soils. Clayey soils (with gypsum) and light loams are commonly
used.​

●​ Special needs: Boron and zinc are critical for cotton. Deficiencies cause poor boll set.
Foliar B and Zn sprays (e.g. 0.5–1 g/L) are recommended if
[Link]. Apply FYM (5 t/ha) or compost for Ca/Mg supply.​

●​ Adjustment tips: If soil is acidic, lime; if too alkaline, gypsum. Low N: side-dress urea
in two splits. Low P or K: use DAP/SSP or MOP in early growth. Micronutrient sprays
(B, Zn, Mg) if deficiencies appear.​

Sugarcane
●​ Soil pH: 6.0–7.5 (slightly acidic to neutral) (optimal)[Link]. Tolerates pH up to
8.0 but extreme pH (>8) reduces nutrient uptake.​

●​ N–P–K: ~130:52:40 kg/[Link] (≈320:130:100 kg/ha). Typical rates are ~130 kg

urea, 52 kg DAP, 40 kg MOP per acre [Link]. Split urea (½ at planting, ½ at
tillering) improves efficiency.​

●​ Soil moisture: Very high – sugarcane is a water-demanding crop. Requires plentiful

irrigation/rain (150–200 cm water per season). Keep soil moist but not waterlogged (good
drainage critical). Frequent irrigation (5–7 days in hot weather) is [Link].​
 ●​ Soil type: Deep, well-drained loamy to sandy-loam soils. Alluvial and red loams are
excellent. Soils should be deep (>60 cm) and fertile; avoid shallow or poorly drained
sodic soils.​

●​ Special needs: Sulfur and zinc are often deficient. Apply gypsum (200 kg/acre) at
planting for S and gypsum’s [Link]. Add ZnSO₄·7H₂O (~5 kg/acre) for
[Link]. Also apply Mg (25 kg Epsom salt/acre) and a starter mix as
[Link].​

●​ Adjustment tips: On acidic soils apply lime; on saline/alkaline soils apply gypsum.
Ensure adequate FYM (5–10 t/ha) for organic matter. If N is low, topdress urea; if P low,
apply DAP; if K low, MOP. Correct Zn/S as above, and iron chelates for chlorosis if
needed.​

Chili (Capsicum)
●​ Soil pH: 6.5–[Link]. Slightly acidic to neutral soils optimize
nutrient uptake.​

●​ N–P–K: Approximately 160–200:50–70:100–150 kg/[Link]. Studies suggest

optimal ranges of ~160–197 N, 42–72 P₂O₅, and 90–168 K₂O per ha for maximum
[Link]. A common regimen might be ~150 kg N, 60 kg P₂O₅, 120 kg K₂O per
ha (split doses).​

●​ Soil moisture: Even, adequate moisture; avoid stress. Drip or careful irrigation (70–80%
field capacity) helps fruit set. Do not let soil dry out during flowering.​

●​ Soil type: Light-to-medium loams with good drainage. Chili also grows on sandy loam if
irrigated; heavy clays are less suitable due to waterlogging risk.​

●​ Special needs: High calcium and boron improve fruit quality. Blossom-end rot (caused
by Ca deficiency) can be prevented by calcium nitrate sprays. Micronutrients: foliar Zn or
B if deficiency symptoms occur. Regular FYM (5–10 t/ha) ensures organic matter and
balanced nutrients.​

●​ Adjustment tips: Apply lime if soil is acidic; elemental S if alkaline. For low N,
side-dress urea in splits; if P low, use DAP; if K low, apply MOP. Foliar feed
 micronutrients as needed.​

Tomato
●​ Soil pH: 6.0–[Link]. Neutral soils (pH≈6.5) ensure availability of Fe,
Mn, Zn.​

●​ N–P–K: ~200:250:250 kg/[Link]. Hybrid tomatoes often use very high

nutrient rates (TNAU recommends ~200 kg N, 250 kg P₂O₅, 250 kg K₂O per
ha)[Link]. Split applications (basal + fertigation) are common.​

●​ Soil moisture: Consistently moist (drip irrigation preferred). Maintain ≈70–80% field
capacity; water stress at flowering/fruiting causes fruit drop. Avoid waterlogging by
raised beds.​

●​ Soil type: Sandy loam to clay [Link]. Tomatoes can grow in a wide
range, but well-drained fertile loams are best. Heavy soils are OK if drainage is good.​

●​ Special needs: High calcium demand (prevents blossom rot) – apply calcium nitrate or
gypsum. Also require magnesium and micronutrients (iron, zinc,
boron)[Link]. Foliar sprays of Zn or B (e.g. 0.1% solutions) at flowering can
boost yield/quality. Incorporate 25 t/ha FYM at planting to improve structure and
slow-release [Link].​

●​ Adjustment tips: Lime acidic soils; elemental S/gypsum if too alkaline. Ensure adequate
FYM or compost. If N is low, apply urea in splits; if P low, use DAP; if K low, apply
MOP or sulfate of potash. For micronutrient deficiencies (yellowing veins, poor fruit set),
spray ZnSO₄ or borax.​

Soybean
 ●​ Soil pH: 6.5–[Link]. Neutral soils favor nodulation; avoid acidic
soils (<6.0).​

●​ N–P–K: No N fertilizer if properly inoculated (plants fix N). Typical P₂O₅ needs
~20–40 kg/ha and K₂O ~20–40 kg/ha. Adequate P is especially important for early
growth. Apply SSP or DAP (25–30 kg P₂O₅/ha) at sowing.​

●​ Soil moisture: Moderate – soy is somewhat drought-tolerant but needs moisture at

flowering. Irrigate/rainfall of ~500–600 mm/season is sufficient. Good drainage is
needed; waterlogging kills soybean.​

●​ Soil type: Deep loamy and clay loam soils are preferred. Black cotton soils and red loams
are widely grown [Link]. Poor, sandy soils give lower yields.​

●​ Special needs: Sulfur and zinc are commonly deficient in soybean

[Link]. Apply gypsum (e.g. 200 kg/ha to supply ~30 kg S)[Link]
and ZnSO₄·7H₂O (~25 kg/ha) at sowing. Ensure seed is [Link]
for full N fixation. Molybdenum (Mo) seed coating (5–10 g/ha) may be needed on acidic
soils.​

●​ Adjustment tips: If soil is acidic, lime to pH 6.5; if alkaline, gypsum. Add P and K
fertilizers if tests show deficiency. Maintain organic matter (legume residues or FYM) to
support fixation. Use Rhizobium inoculants every [Link]​

Pulses (e.g. chickpea, pigeon pea, lentil)

●​ Soil pH: 6.0–7.5 (neutral to slightly alkaline)[Link]. Very acidic soils
impede rhizobia.​

●​ N–P–K: No N fertilizer if inoculated (legumes fix N). Apply ~25–30 kg P₂O₅/ha

(SSP/DAP) and 25–30 kg K₂O/ha (MOP) basally. Pulses generally need more P and K for
good nodulation and pod fill.​

●​ Soil moisture: Moderate, often rainfed. Good soil moisture at flowering and pod-fill is
critical. Irrigate once or twice if dry (before flowering and at pod set); avoid excess
moisture.​
●​ Soil type: Well-drained loam/sandy loam soils. Pulses dislike waterlogging. Deep, fertile
soils give better yields; light soils with good moisture can work for chickpea.​

●​ Special needs: Rhizobium inoculation for N-fixation (even if planted before) is essential.
Boron deficiency is common (especially chickpea) – apply 1–2 kg/ha borax at sowing.
Molybdenum is critical for all legumes – use Mo-enriched fertilizers or foliar Mo (e.g.
Na₂MoO₄). Many pulses benefit from zinc (foliar ZnSO₄ if leaves pale).​

●​ Improvement tips: Correct low pH with lime (pH <6.5); use gypsum on alkaline soils. If
P is low, side-apply SSP/DAP. If K is low, apply MOP. Add farmyard manure or compost
yearly to improve structure. Rotate pulses with cereals to maintain soil N and break
disease cycles.