SAM 2++: Enhanced Tracking & Segmentation

• SAM 2++ is a family of enhancements to SAM 2 that refines prompt construction, memory matching, and decoder design for robust tracking and segmentation.
• The approach integrates automated proxy prompts, task-adaptive memory, and specialized decoders, addressing challenges in biomedical imaging and video tracking.
• Empirical results show state-of-the-art improvements in benchmarks with significant gains in IoU, HOTA, Dice, and related performance metrics.

SAM 2++ denotes a line of research that extends the Segment Anything Model 2 (SAM 2) beyond promptable video object segmentation toward more robust tracking and domain-adapted segmentation. In the current literature, the term has two coexisting uses. First, it functions as an umbrella label for enhanced SAM 2 systems that add automated prompting, specialized memory, biomedical adaptation, or tracking refinements. Second, it names a specific unified model for tracking masks, boxes, and points with task-specific prompt encoding, a unified decoder, task-adaptive memory, and a dedicated tri-granular dataset (Zhang et al., 21 Oct 2025). The dual usage is consequential: SAM 2++ is not yet a single standardized architecture, but a family of designs centered on improving prompt formation, memory matching, temporal stability, and adaptation to context-dependent concepts (Zhao et al., 2024).

1. Terminological scope and historical usage

The expression “SAM 2++” emerged as a shorthand for improvements over vanilla SAM 2 in settings where the base model exhibits prompt sensitivity, temporal drift, weak context modeling, or domain mismatch. In this broader sense, the label has been attached to biomedical adaptation, automated prompting for medical segmentation, and memory-stabilized ultrasound tracking, while a later work formalized it as the title of a unified tracking model (Xinyi et al., 5 Feb 2025, Yan et al., 2024, Dialameh et al., 21 Oct 2025, Zhang et al., 21 Oct 2025).

Variant	Primary modification	Domain
Proxy Prompt	auto-generated proxy prompt from non-target image-mask pairs	few-shot medical segmentation
BioSAM 2	frozen prompt encoder with fine-tuned image encoder and mask decoder	biomedical images and videos
EMA-SAM	confidence-weighted exponential moving average pointer in the memory bank	PTMC segmentation in ultrasound videos
SAM 2++	task-specific prompts, unified decoder, task-adaptive memory	tracking at any granularity

A common misconception is that the name refers only to the unified tracker introduced in late 2025. The literature instead shows a broader pattern in which “SAM 2++” can denote either a specific architecture or a research agenda for strengthening SAM 2 under harder regimes, especially context-dependent concepts, medical imaging, and long-term video tracking (Zhao et al., 2024).

2. Core architectural motif: prompt, memory, and decoder augmentation

Across the literature, SAM 2 is treated as a modular backbone with an image encoder, prompt encoder, memory mechanism, and promptable decoder. One explicit recap describes SAM-2 as comprising a frozen hierarchical ViT image encoder, a memory-augmented transformer, a promptable decoder that outputs a binary mask, an object-pointer token, and an occlusion score, and a fixed-capacity FIFO memory bank of spatial features and pointer tokens (Dialameh et al., 21 Oct 2025). Enhancements marketed or discussed as SAM 2++ almost always intervene at one of four loci.

The first locus is prompt construction. Some works replace manual clicks or boxes with structured prompts derived from context, exemplars, or alternate granularities. Proxy Prompt replaces direct target prompting with contextual embeddings extracted from non-target image-mask pairs (Xinyi et al., 5 Feb 2025). The formal SAM 2++ tracker encodes mask prompts, box prompts, and point prompts into a shared prompt space, with point tracking additionally using a Gaussian mask (Zhang et al., 21 Oct 2025).

The second locus is memory design. Memory is the dominant intervention point in video and volumetric settings. EMA-SAM inserts a persistent EMA pointer that is never evicted from the bank (Dialameh et al., 21 Oct 2025). SLM-SAM 2 splits memory into short-term and long-term banks with separate attention modules (Chen et al., 3 May 2025). HiM2SAM distinguishes short-term and long-term memory frames under motion- and distractor-aware criteria (Chen et al., 10 Jul 2025). CamSAM2 adds object prototypes distilled from high-resolution masked features of previous frames (Zhou et al., 25 Mar 2025).

The third locus is decoder specialization. BioSAM 2 simplifies the mask decoder to output exactly one mask per frame or slice, dropping the mask-ranking head and replacing bidirectional cross-attention with single-mask transformer layers (Yan et al., 2024). The unified SAM 2++ tracker retains a two-way-transformer style decoder but adds heads for masks, boxes, and points (Zhang et al., 21 Oct 2025). CamSAM2 appends a learnable decamouflaged token to the decoder without modifying SAM2’s original weights (Zhou et al., 25 Mar 2025).

The fourth locus is adaptation regime. The spectrum ranges from trainless augmentation to full joint training. HiM2SAM is explicitly trainless (Chen et al., 10 Jul 2025). Proxy Prompt fine-tunes only LoRA adapters while keeping the original SAM encoder and decoder otherwise frozen (Xinyi et al., 5 Feb 2025). BioSAM 2 fine-tunes the image encoder and mask decoder while freezing the prompt encoder (Yan et al., 2024). The formal SAM 2++ model is jointly trained across video object segmentation, single-object tracking, and point tracking with task-specific memory modules (Zhang et al., 21 Oct 2025). This suggests a shared engineering view of SAM 2 as a prompt-memory-decoder substrate rather than a fixed end model.

3. The formal unified model: tracking masks, boxes, and points

The paper “SAM 2++: Tracking Anything at Any Granularity” defines the most specific use of the term. Its objective is to unify three tasks that differ only in target granularity: mask-level video object segmentation, bounding-box single-object tracking, and point tracking (Zhang et al., 21 Oct 2025). The central claim is that all three can be cast within a memory-matching framework if prompts, decoder outputs, and memory representation are made task-adaptive rather than task-isolated.

At the input side, SAM 2++ uses a shared prompt encoder with task-specific embeddings. Mask prompts are dense binary masks, box prompts are two corner points, and point prompts are a 2D coordinate plus a Gaussian mask. In the formulation reported for the model,

$$\mathcal P_{dense}= \begin{cases} \mathrm{Conv}_{dense}(I_{mask}), & \text{mask prompt},\ \mathrm{Conv}_{dense}(G_{point}), & \text{point prompt},\ 0, & \text{otherwise}. \end{cases}$$

The Gaussian mask is annealed from coarse to fine during training, so point localization is first broad and later precise (Zhang et al., 21 Oct 2025).

At the output side, the model extends SAM 2’s mask decoder into a unified decoder. Memory-conditioned image features, sparse prompt tokens, dense prompt features, mask tokens, an object-existence token, and an IoU-prediction token are fused by a Two-Way Transformer. The resulting token set can produce candidate masks, boxes via a corner head, and points via soft-argmax during training or hard argmax at inference. Candidate quality is scored by the IoU token, and the best candidate is selected for the task (Zhang et al., 21 Oct 2025).

The memory mechanism is task-adaptive rather than fully shared. For each granularity $\rho\in\{\text{mask, box, point}\}$, the model uses a dedicated memory encoder and decoupled LoRA parameters inside memory attention:

$$\bar{\bar F}_{img,\rho}=\mathrm{MemEn}_\rho(F_{img},M^*_\rho),\qquad \mathcal{MB}_\rho=\mathrm{FIFO}\bigl(\mathcal{MB}_\rho\cup\{\bar{\bar F}_{img,\rho},\varepsilon_{pointer,\rho}\}\bigr),\qquad \bar F_{img}=\mathrm{MemAttn}_\rho(F_{img},\mathcal{MB}_\rho).$$

The rationale is that mask tracking requires fine pixel-level memory, box tracking requires coarse localization, and point tracking requires tightly localized memory around a Gaussian prompt (Zhang et al., 21 Oct 2025).

Training is organized around a customized data engine and the Tracking-Any-Granularity (TAG) dataset. TAG contains 6 000 videos and 2.2 M frames, with 2.15 M segmentation masks, 2.15 M bounding boxes, and 2.64 M key-points; it is split into 150 val + 150 test videos (Zhang et al., 21 Oct 2025). A single batch comes from exactly one task, sampled with probability approximately 0.1:0.4:0.5 for mask:box:point. Each training sequence has 8 frames, 1–2 conditional frames, and up to 7 corrective clicks. Losses combine focal and Dice mask supervision, IoU regression, object-existence classification, and either box or point regression depending on the task (Zhang et al., 21 Oct 2025).

Empirically, the model is reported to set a new state of the art across multiple benchmarks. On video object segmentation it achieves 66.4 HOTA on BURST, 82.2 / 78.7 / 85.7 on LVOS, 74.6 / 70.6 / 78.6 on MOSE, and 87.4 / 84.2 / 90.7 on TAG-VOS. On single-object tracking it reports 80.7 AO on GOT-10k, 78.0 AUC on TAG-SOT, and 86.0 AUC on TrackingNet. On point tracking it reaches 72.9 PCK-T on BADJA, 66.2 AJ on TAG, and 37.7 AJ on TAG-PT (Zhang et al., 21 Oct 2025). The associated ablations indicate that decoupling only the memory encoders and LoRA adapters per task restores performance lost under naive parameter sharing, and that joint training across granularities yields positive transfer rather than task interference (Zhang et al., 21 Oct 2025).

4. Biomedical and medical reinterpretations

Before the formal tri-granularity tracker, several biomedical papers used “SAM 2++” to describe domain-specific extensions of SAM 2. These systems are unified less by a canonical architecture than by a shared attempt to overcome weak prompting, poor low-level adaptation, or unstable propagation in medical imagery.

Proxy Prompt treats SAM and SAM 2 as promptable backbones and replaces manual target prompting with an auto-generated contextual prompt derived from non-target image-mask pairs (Xinyi et al., 5 Feb 2025). Its 3-step context-selection strategy begins with Vision Mamba encoding of concatenated support image-mask pairs, continues with a Bridge Unit that concatenates frozen SAM encoder features with Mamba outputs and applies CBAM plus residual blocks, and ends with a Selective Map that computes target relevance. The Selective Map is defined as

$$\mathrm{Selective}=\bigl[2(\hat F_{\rm sup}^T\hat F_x)-\hat F_{\rm sup}^2\bigr]/\sqrt{C},$$

after which

$$E_{\rm ctx}=A_{\rm agg}\cdot \mathrm{Softmax}(\mathrm{Selective}).$$

A Contextual Colorization Module with $L=4$ identical blocks then alternates feature-to-context and reversed context-to-feature cross-attention, sharpening channels aligned with user-defined mask regions and amplifying distinctive edges and textures (Xinyi et al., 5 Feb 2025). In the SAM 2 case, the resulting prompt is injected into memory attention as high-dimensional prompt vectors. The method fine-tunes only LoRA adapters, uses $M=16$ training support pairs per modality, and reports few-shot Dice scores of 88.2% for REFUGE2 Disc, 85.6% for REFUGE2 Cup, 78.9% for STARE Vessel, 81.5% for FPA-PS, and 88.0% for FPA-FH, for an overall average of 84.4%. On the JNU-IFM video benchmark it improves over MedSAM 2 by +17.5 pp and over SAM 2-box by +2.3 pp, with $\sigma(\mathrm{Dice})\le 1.0\%$ across 5 random support pairs, compared with 29.1% for MedSAM 2 (Xinyi et al., 5 Feb 2025).

BioSAM 2 adapts SAM 2 to biomedical images and videos by freezing the prompt encoder and fine-tuning only the image encoder and mask decoder (Yan et al., 2024). It uses the tiny Hiera transformer with 12 transformer blocks and 384-dim embeddings, applies layer decay, keeps positional embeddings unchanged, and simplifies the decoder to output exactly one mask per frame or volume slice, dropping the mask-ranking head while retaining the auxiliary occlusion head (Yan et al., 2024). Training uses point prompts only, with 1–5 automatically sampled clicks, Dice plus BCE loss, AdamW with initial learning rate $1e{-4}$, and cosine decay after warm-up (Yan et al., 2024). On EndoVis 2017 Instruments, BioSAM 2 reports 0.6251±0.2897 DSC and 0.6427±0.3095 NSD; on the NeurIPS 2022 Cell Segmentation Challenge, it reports 0.5792±0.2666 F1 and 0.7436±0.2104 NSD. Relative to the best zero-shot SAM 2 baseline with 5 clicks, the gains are +0.087 on DSC and +0.090 on F1 (Yan et al., 2024). The same source notes that BioSAM 2 video results were not yet reported.

SLM-SAM 2 addresses volumetric medical annotation by replacing SAM 2’s single memory-attention branch with a dual-branch short-long memory module (Chen et al., 3 May 2025). It preserves the image encoder, prompt encoder, and mask decoder, but conditions the current slice on a short-term bank $\mathcal M_{short}$ and a long-term bank $\rho\in\{\text{mask, box, point}\}$0 with separate attention modules and a lightweight fusion module. The short-term bank has capacity $\rho\in\{\text{mask, box, point}\}$1, while the long-term bank keeps $\rho\in\{\text{mask, box, point}\}$2 recent plus $\rho\in\{\text{mask, box, point}\}$3 prompted encoding (Chen et al., 3 May 2025). This design targets boundary over-propagation, especially when the target is present in a previous slice but absent in the current one. In the 5-Volume setting it reports an average DSC gain of approximately +0.14 over SAM 2, and in the 1-Volume setting an average +0.11 gain. The key ablation gives 0.69 DSC for the SAM 2 baseline, 0.78 for a recent-only bank of size 1, and 0.83 for the full short-long memory configuration (Chen et al., 3 May 2025). On target-absent slices, SLM-SAM 2 maintains near-perfect DSC (≈1.0), whereas SAM 2’s DSC drops to zero under complete over-segmentation (Chen et al., 3 May 2025).

Taken together, these biomedical variants show that “SAM 2++” in medicine usually refers not to a broader general tracker, but to methods that re-engineer prompting, decoder specialization, or propagation while keeping the SAM 2 backbone recognizable.

5. Memory-centric and scene-specific video extensions

A second major branch of SAM 2++ research emphasizes temporal robustness under occlusion, drift, camouflage, and long-term reappearance. These works often aim for minimal perturbation of SAM 2 while modifying the memory pathway or adding small task-specific modules.

EMA-SAM inserts a persistent EMA pointer into SAM-2’s memory bank (Dialameh et al., 21 Oct 2025). If $\rho\in\{\text{mask, box, point}\}$4 is the instantaneous pointer token and $\rho\in\{\text{mask, box, point}\}$5 is visibility confidence, the update rule is

$\rho\in\{\text{mask, box, point}\}$6

with $\rho\in\{\text{mask, box, point}\}$7 and amplification $\rho\in\{\text{mask, box, point}\}$8 for cross-attention (Dialameh et al., 21 Oct 2025). High confidence makes the prototype quickly track the current pointer; low confidence makes it freeze. The method adds only one small MLP (∼0.01 M parameters) plus a vector update and normalization, for total additional FLOPs < 0.1 % of SAM-2’s baseline ∼200 GFLOPs per frame, while preserving ≈30 FPS on a single A100 (Dialameh et al., 21 Oct 2025). On the PTMC-RFA dataset it improves maxDice from 0.82 to 0.86, maxIoU from 0.72 to 0.76, and reduces false positives from ≈0.14 to 0.10 (–29%). On external benchmarks it raises CVC-300-TV maxDice from 0.906 to 0.924, CVC-612-V from 0.901 to 0.907, and keeps CVC-612-T maxDice at 0.880 while improving maxIoU from 0.820 to 0.833 (Dialameh et al., 21 Oct 2025).

CamSAM2 targets video camouflaged object segmentation without modifying SAM2’s original weights (Zhou et al., 25 Mar 2025). It adds a learnable decamouflaged token $\rho\in\{\text{mask, box, point}\}$9, Implicit Object-Aware Fusion (IOF) for combining early-layer features with memory-conditioned features, Explicit Object-Aware Fusion (EOF) for refinement using current mask logits and temporal prototypes, and Object Prototype Generation (OPG) to summarize masked high-resolution features into a small set of prototypes via farthest point sampling and one iteration of k-means (Zhou et al., 25 Mar 2025). The final logits are modulated by the decamouflaged token output:

$$\bar{\bar F}_{img,\rho}=\mathrm{MemEn}_\rho(F_{img},M^*_\rho),\qquad \mathcal{MB}_\rho=\mathrm{FIFO}\bigl(\mathcal{MB}_\rho\cup\{\bar{\bar F}_{img,\rho},\varepsilon_{pointer,\rho}\}\bigr),\qquad \bar F_{img}=\mathrm{MemAttn}_\rho(F_{img},\mathcal{MB}_\rho).$$0

The full model adds only 0.5 M parameters and reports large gains on three VCOS benchmarks. On MoCA-Mask with 1-click prompt and Hiera-T, mDice rises from 52.1 to 64.3 and mIoU from 44.8 to 54.6. On SUN-SEG-Hard with mask prompt, mDice rises from 61.0 to 80.6, a +19.6 gain (Zhou et al., 25 Mar 2025). The ablation on MoCA-Mask shows progressive gains from the decamouflaged token, IOF, EOF, and OPG, with the full model reaching 64.3 mDice / 54.6 mIoU (Zhou et al., 25 Mar 2025).

HiM2SAM pursues long-term tracking through a trainless combination of hierarchical motion estimation and memory-bank optimization (Chen et al., 10 Jul 2025). The first motion stage is a Kalman filter over box center, velocity, and scale. The second stage is a selective point-tracker refinement using CoTracker3, invoked only when the combined confidence falls below a threshold (Chen et al., 10 Jul 2025). Memory is split into short-term and long-term pools; frames enter short-term memory when both motion and mask-overlap criteria are met, and enter long-term memory when proposal ambiguity measured by directed Hausdorff distance exceeds a threshold (Chen et al., 10 Jul 2025). On LaSOT, the large model improves AUC from 68.5 to 75.1, a +9.6% relative gain; on LaSOT$$\bar{\bar F}_{img,\rho}=\mathrm{MemEn}_\rho(F_{img},M^*_\rho),\qquad \mathcal{MB}_\rho=\mathrm{FIFO}\bigl(\mathcal{MB}_\rho\cup\{\bar{\bar F}_{img,\rho},\varepsilon_{pointer,\rho}\}\bigr),\qquad \bar F_{img}=\mathrm{MemAttn}_\rho(F_{img},\mathcal{MB}_\rho).$$1 it improves from 58.6 to 62.8, a +7.2% relative gain (Chen et al., 10 Jul 2025). The ablation on LaSOT shows 68.32 AUC for baseline SAM2, 70.81 with Kalman filter only, 72.67 with short-term memory only, 73.71 with long-term memory only, and 75.09 for the full system, at +3.68 ms latency per frame (Chen et al., 10 Jul 2025).

These variants share a narrow technical premise: substantial gains can be obtained by reweighting or restructuring memory, rather than by replacing the SAM 2 backbone. This is particularly clear in EMA-SAM, CamSAM2, and HiM2SAM, which all preserve most of the original architecture and intervene through lightweight tokens, prototypes, or motion-aware memory rules.

6. Evaluation findings, misconceptions, and research trajectory

The most systematic precursor to the SAM 2++ agenda is a comprehensive evaluation of SAM and SAM 2 on 11 context-dependent concepts across 2D and 3D images and videos in natural, medical, and industrial scenes (Zhao et al., 2024). The study shows that SAMs remain strong on context-independent categories, with the original SAM paper reporting mIoU in the mid-80 percent range on COCO Panoptic validation for rigid classes such as people, cars, and roads. By contrast, performance on context-dependent concepts is far less stable. For Salient Object Detection, SAM (box) reaches $$\bar{\bar F}_{img,\rho}=\mathrm{MemEn}_\rho(F_{img},M^*_\rho),\qquad \mathcal{MB}_\rho=\mathrm{FIFO}\bigl(\mathcal{MB}_\rho\cup\{\bar{\bar F}_{img,\rho},\varepsilon_{pointer,\rho}\}\bigr),\qquad \bar F_{img}=\mathrm{MemAttn}_\rho(F_{img},\mathcal{MB}_\rho).$$2, $$\bar{\bar F}_{img,\rho}=\mathrm{MemEn}_\rho(F_{img},M^*_\rho),\qquad \mathcal{MB}_\rho=\mathrm{FIFO}\bigl(\mathcal{MB}_\rho\cup\{\bar{\bar F}_{img,\rho},\varepsilon_{pointer,\rho}\}\bigr),\qquad \bar F_{img}=\mathrm{MemAttn}_\rho(F_{img},\mathcal{MB}_\rho).$$3, and MAE $$\bar{\bar F}_{img,\rho}=\mathrm{MemEn}_\rho(F_{img},M^*_\rho),\qquad \mathcal{MB}_\rho=\mathrm{FIFO}\bigl(\mathcal{MB}_\rho\cup\{\bar{\bar F}_{img,\rho},\varepsilon_{pointer,\rho}\}\bigr),\qquad \bar F_{img}=\mathrm{MemAttn}_\rho(F_{img},\mathcal{MB}_\rho).$$4, while SAM 2 (box) is slightly weaker at $$\bar{\bar F}_{img,\rho}=\mathrm{MemEn}_\rho(F_{img},M^*_\rho),\qquad \mathcal{MB}_\rho=\mathrm{FIFO}\bigl(\mathcal{MB}_\rho\cup\{\bar{\bar F}_{img,\rho},\varepsilon_{pointer,\rho}\}\bigr),\qquad \bar F_{img}=\mathrm{MemAttn}_\rho(F_{img},\mathcal{MB}_\rho).$$5, $$\bar{\bar F}_{img,\rho}=\mathrm{MemEn}_\rho(F_{img},M^*_\rho),\qquad \mathcal{MB}_\rho=\mathrm{FIFO}\bigl(\mathcal{MB}_\rho\cup\{\bar{\bar F}_{img,\rho},\varepsilon_{pointer,\rho}\}\bigr),\qquad \bar F_{img}=\mathrm{MemAttn}_\rho(F_{img},\mathcal{MB}_\rho).$$6, and MAE $$\bar{\bar F}_{img,\rho}=\mathrm{MemEn}_\rho(F_{img},M^*_\rho),\qquad \mathcal{MB}_\rho=\mathrm{FIFO}\bigl(\mathcal{MB}_\rho\cup\{\bar{\bar F}_{img,\rho},\varepsilon_{pointer,\rho}\}\bigr),\qquad \bar F_{img}=\mathrm{MemAttn}_\rho(F_{img},\mathcal{MB}_\rho).$$7. Automatic “everything” prompts drop to $$\bar{\bar F}_{img,\rho}=\mathrm{MemEn}_\rho(F_{img},M^*_\rho),\qquad \mathcal{MB}_\rho=\mathrm{FIFO}\bigl(\mathcal{MB}_\rho\cup\{\bar{\bar F}_{img,\rho},\varepsilon_{pointer,\rho}\}\bigr),\qquad \bar F_{img}=\mathrm{MemAttn}_\rho(F_{img},\mathcal{MB}_\rho).$$8 (Zhao et al., 2024). On 2D medical lesion segmentation, SAM box achieves approximately 0.70 IoU / 0.82 Dice, point prompting about 0.65 / 0.78, and automatic prompting about 0.50 / 0.65 (Zhao et al., 2024). On BraTS2020 3D brain tumors, SAM 2 (5 frames) reaches IoU ≈ 0.82 and Dice ≈ 0.88, surpassing a 3D U-Net baseline (Zhao et al., 2024).

The same evaluation isolates prompt sensitivity as a central weakness. Perturbations of ±10 px in point coordinates, ±10% box-edge jitter, or 5-step mask erosion or dilation can produce 5–10 pp relative drops in IoU or F-measure (Zhao et al., 2024). The reported relative change

$$\bar{\bar F}_{img,\rho}=\mathrm{MemEn}_\rho(F_{img},M^*_\rho),\qquad \mathcal{MB}_\rho=\mathrm{FIFO}\bigl(\mathcal{MB}_\rho\cup\{\bar{\bar F}_{img,\rho},\varepsilon_{pointer,\rho}\}\bigr),\qquad \bar F_{img}=\mathrm{MemAttn}_\rho(F_{img},\mathcal{MB}_\rho).$$9

has standard deviation across five trials of $$\mathrm{Selective}=\bigl[2(\hat F_{\rm sup}^T\hat F_x)-\hat F_{\rm sup}^2\bigr]/\sqrt{C},$$0, but mean $$\mathrm{Selective}=\bigl[2(\hat F_{\rm sup}^T\hat F_x)-\hat F_{\rm sup}^2\bigr]/\sqrt{C},$$1 is often −0.05 to −0.10, indicating systematic rather than merely noisy degradation (Zhao et al., 2024). A second misconception is therefore that SAM 2++ work is only about scaling or domain-specific data; much of it is an attempt to mitigate brittle prompt-response behavior.

The same source also reports that in-context learning in SAM 2 can already recover part of the gap. Supplying $$\mathrm{Selective}=\bigl[2(\hat F_{\rm sup}^T\hat F_x)-\hat F_{\rm sup}^2\bigr]/\sqrt{C},$$2 exemplar image-mask pairs before inference yields 5–8 pp IoU gain on COD, TOS, SD, and four LOS tasks, with IoU increasing from ∼0.60 to ∼0.68 (Zhao et al., 2024). This finding anticipates later exemplar-driven or context-driven methods such as Proxy Prompt, and suggests that memory conditioning on support data is a general route to SAM 2++ behavior.

The future directions proposed across the literature are consistent. The evaluation paper recommends hybrid multi-stage prompt embedding, memory encoders for arbitrary exemplar modalities, cross-prompt attention heads, joint CI–CD pretraining on a unified ConceptNet, concept-balanced sampling with

$$\mathrm{Selective}=\bigl[2(\hat F_{\rm sup}^T\hat F_x)-\hat F_{\rm sup}^2\bigr]/\sqrt{C},$$3

prompt-noise augmentation, multi-modal prompts, and dual-branch prompt decoders (Zhao et al., 2024). Biomedical work proposes semantic heads, text/image co-grounding, learnable cold-start modules, PETL/LoRA-style adapters, and 3D-aware attention or 3D conv layers (Yan et al., 2024). EMA-SAM further suggests multi-object or multi-class tracking by maintaining separate EMA pointers, adaptive gain schedules, weakly supervised or semi-supervised EMA update, and cross-modal integration such as Doppler or elasticity (Dialameh et al., 21 Oct 2025).

A plausible implication is that future uses of “SAM 2++” will continue to span two levels simultaneously: a specific unified tracking model on one hand, and a broader methodological program for making SAM 2 less prompt-fragile, more memory-stable, and more adaptable to context-dependent or domain-shifted regimes on the other.