Image Classification on Brain Tumour MRI
89.8AccuracyHierarchical Spatio-Channel Clustering
Evaluation results
| Method | Links | |
|---|---|---|
| Hierarchical Spatio-Channel ClusteringCompression Budget=6×, Params (M)=33.96, Size (MB)=129.53, FLOPs (G)=1.55, Latency (ms)=1.30, Speed-up=1.38×2026.04 | 89.8 | |
| Hierarchical Spatio-Channel ClusteringParams (M)=34.33, Size (MB)=130.96, FLOPs (G)=2.73, Latency (ms)=1.44, Speed-up=1.26×, Compression=3×2026.04 | 87.96 | |
| BaselineCompression Budget=6×, Params (M)=35.83, Size (MB)=136.68, FLOPs (G)=8.21, Latency (ms)=1.80, Speed-up=1.00×2026.04 | 87.76 | |
| BaselineParams (M)=35.83, Size (MB)=136.68, FLOPs (G)=8.21, Latency (ms)=1.80, Speed-up=1.00×, Compression=3×2026.04 | 87.76 | |
| Global SVDParams (M)=34.33, Size (MB)=130.96, FLOPs (G)=2.82, Latency (ms)=1.49, Speed-up=1.21×, Compression=3×2026.04 | 87.76 | |
| TuckerCompression Budget=6×, Params (M)=33.95, Size (MB)=129.52, FLOPs (G)=1.46, Latency (ms)=1.11, Speed-up=1.62×2026.04 | 85.31 | |
| Global SVDCompression Budget=6×, Params (M)=33.96, Size (MB)=129.53, FLOPs (G)=1.47, Latency (ms)=1.21, Speed-up=1.49×2026.04 | 84.9 | |
| TuckerParams (M)=34.32, Size (MB)=130.93, FLOPs (G)=2.79, Latency (ms)=1.43, Speed-up=1.26×, Compression=3×2026.04 | 84.69 | |
| Lingshu 7BMedical Samples=5M2026.06 | 78.11 | |
| Qwen2.5-VL-7B + OPENMEDREASONMedical Samples=450K2026.06 | 75.35 | |
| OctoMed 7BMedical Samples=8M2026.06 | 71.42 | |
| QoQ-Med-VL 7BMedical Samples=2.6M2026.06 | 62.22 | |
| MedGemma 1.5 4BMedical Samples=30M2026.06 | 41.67 | |
| MedGemma 27BMedical Samples=30M2026.06 | 40.55 | |
| Qwen2.5-VL-7BMedical Samples=N/A2026.06 | 36 | |
| MedVL-Thinker 7BMedical Samples=200K2026.06 | 30.91 |