Abstract: Free surfaces serve as critical channels for blast energy release, and their quantity directly governs the evolution of explosion-induced fracture fields and rock fragmentation efficiency, which exerts a significant influence on mining blasting performance and safe production. To reveal the mechanism of the number of free surfaces on blast energy evolution and fragment size distribution, this study employs a digital laser dynamic caustics experimental system to conduct blast loading tests under different numbers of free surfaces. Fractal theory and a MATLAB box-counting program are combined to quantify the characteristics of fracture networks, and an intelligent identification system for blasting fragmentation is developed based on the SAM deep learning model. Field verification tests of side-hole pre-split blasting are performed in the thin ore vein of the Gongchangling underground iron mine.The results show that the number of free surfaces is positively correlated with crack propagation velocity, dynamic stress intensity factor, and fractal dimension. Specimens with zero free surfaces exhibit sparse and disordered cracks, whereas those with four free surfaces form a fully penetrated and dense fracture network. The peak crack propagation velocity increases by 38.6%, and the peak dynamic stress intensity factor rises by 71.4%. The fractal dimension increases from 1.1050 to 1.4196 with the increase in free surfaces, indicating a significant improvement in fracture complexity. Dual-energy pre-split blasting of side holes can actively create controllable blasting free surfaces and optimize energy release paths, reducing the boulder yield from 19.3% to 4.2% and the non-uniformity coefficient by 48.6%, leading to a more uniform fragment size distribution.