• Chatti. Subbalakshmi
• Dr.B.Srinivasa Rao
• Mr.Chamakuri Upendar

While Vision Transformers (ViTs) have redefined the state-of-the-art in semantic segmentation, their quadratic computational complexity O(N2) remains a barrier for edge deployment. We propose a multi-stage compression framework that integrates Dynamic Structural Pruning (DSP) and 8-bit Quantization-Aware Training (QAT). Our approach reduces the energy footprint by 65% while maintaining a Mean Intersection over Union (mIoU) within 1.2% of the baseline on the Cityscapes dataset. We demonstrate real-time inference (≥ 30FPS) on an NVIDIA Jetson Orin Nano and Raspberry Pi 5.

Computer Vision, Semantic Segmentation, Machine Learning, Vision Transformers (ViT), Model Compression, Mean Intersection over Union (mIoU).

"Energy-Efficient Transformer Compression for Real-Time Semantic Segmentation on Resource-Constrained Edge Devices", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 1, page no.b444-b447, January-2026, Available :http://www.jetir.org/papers/JETIR2601164.pdf

"Energy-Efficient Transformer Compression for Real-Time Semantic Segmentation on Resource-Constrained Edge Devices", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 1, page no. ppb444-b447, January-2026, Available at : http://www.jetir.org/papers/JETIR2601164.pdf