Pramod Kumar Sahoo

Vegetable production is vital for nutritional security and farm-based livelihoods because vegetables supply essential vitamins, minerals, and dietary diversity while contributing substantially to h...

Manual rice transplanting is labour-intensive, time-consuming, and inconsistent, while conventional root-washed mechanical transplanters using gravity-fed seedling delivery suffer from irregular flow, missed hills, non-uniform spacing, and seedling damage. These limitations necessitate a precisely controlled and experimentally optimized seedling delivery system to enhance transplanting efficiency and uniformity. This study aimed to develop and optimize an electronically controlled transplanting system for root-washed paddy seedlings by integrating a motor-driven crank–rocker mechanism with a Hall sensor-synchronized double-stacked zigzag tray and belt conveyor system, replacing gravity-based feeding with an electronically controlled delivery mechanism. Experiments were designed using Design-Expert 12 under a 3 × 3 factorial arrangement with disc speed (90, 109, and 128 rpm) and belt displacement (10, 15, and 20 mm per stroke) as independent variables. A total of 30 experimental runs were conducted for PB 1692 and PUSA 1612 paddy varieties using seedlings with optimized morphological characteristics (shoot diameter 4 ± 0.2 mm and seedling height 240 ± 5 mm). Responses included seedlings picked, seedlings per hill, missed seedlings, visible damage, and planting efficiency. Quadratic models developed through ANOVA showed significant effects (p < 0.01) of disc speed, belt displacement, and their interaction, with high model accuracy and non-significant lack-of-fit. Multi-response optimization using desirability functions considered 3–4 seedlings per hill, 150–200 seedlings per cycle, minimum misses and damage, and realistic efficiency constraints. Although 90 rpm and 20 mm stroke produced higher seedling output, it caused overfeeding and unrealistic efficiency (> 100%). The optimal condition was identified at approximately 90 rpm and 15 mm per stroke. Under optimized conditions, laboratory performance included 198 ± 0.82 seedlings per 200 strokes, 3.95 ± 0.01 seedlings per hill, 1.25 ± 0.25% misses, 0.125 ± 0.035% visible damage and 98.75 ± 0.25% planting efficiency. Field evaluation of the developed transplanting mechanism in sandy loam soil (48 h sedimentation period) achieved 200 ± 15 mm hill spacing, 3.5 ± 0.5 cm planting depth, and 20 ± 2 hills m ⁻² , with no significant differences between PB 1692 and PUSA 1612. Compared with laboratory results, field missed hills increased by 57–60% due to machine vibration causing seedling misalignment and soil clogging within the finger gap, while mechanical damage increased because inadequate soil anchorage occasionally prevented seedling release, leading to jamming during subsequent strokes. Nevertheless, the sensor-based conveyor reduced missed hills by 50–60% compared with the gravity-assisted system, and the transplanter achieved an effective field capacity of 0.038 ha h ⁻¹ with 50% fewer tray refills and 47% lower transplanting time per hectare.

Brinjal (Solanum melongena L.) is an important vegetable crop worldwide, but its cultivation faces challenges from pests, diseases, and variable environmental conditions that negatively affect quality and yield. Accurate fruit detection in natural field conditions is essential for yield estimation and perception module of automated harvesting, but existing deep learning techniques often require substantial computational support, restricting their deployment on edge devices. This study addresses this gap by evaluating the Faster Objects, More Objects (FOMO) model—a lightweight architecture for resource-constrained platforms—for brinjal detection under diverse field conditions. A dataset of 1500 images was captured under varying illumination and growth stages, annotated using a bounding box-based approach, and used to train FOMO models with 25, 50, and 100 epochs via transfer learning. Post-training quantization to INT8 format was applied to assess improvements in computational efficiency. The Float32 model achieved a precision of 0.827, recall of 0.915, and F1-score of 0.869 at 100 epochs. The INT8-quantized model maintained comparable accuracy (precision 0.829, recall 0.908, F1-score 0.866) while reducing model size by 63.33% (from 0.30 MB to 0.11 MB), inference time by 62.02% (from 65.3 ms to 24.8 ms), and RAM usage by 73.02% (from 887.2 KB to 239.4 KB). These results demonstrate that FOMO combined with INT8 quantization provides an efficient, accurate solution for real-time brinjal detection on edge platforms, supporting the advancement of precision agriculture through intelligent crop monitoring and serving as a perception module for future robotic harvesting systems.