Update docs

Author: benITo47

apps/computer-vision/app/index.tsx

@@ -41,7 +41,6 @@ export default function Home() {
         >
           <Text style={styles.buttonText}>Object Detection Live</Text>
         </TouchableOpacity>
-        <TouchableOpacity style={styles.button}></TouchableOpacity>
         <TouchableOpacity
           style={styles.button}
           onPress={() => router.navigate('instance_segmentation/')}

docs/docs/02-benchmarks/memory-usage.md

@@ -169,17 +169,23 @@ with higher resolutions.
 
 ## Instance Segmentation
 
+:::info
+All the below benchmarks were performed on iPhone 17 Pro (iOS) and OnePlus 12
+(Android).
+:::
+
 :::warning
 Data presented in the following sections is based on inference with forward_640 method.
 :::
 
-| Model       | Android (XNNPACK) [MB] | iOS (XNNPACK) [MB] |
-| ----------- | ---------------------- | ------------------ |
-| YOLO26N_SEG | 92                     | 668                |
-| YOLO26S_SEG | 220                    | 712                |
-| YOLO26M_SEG | 570                    | 815                |
-| YOLO26L_SEG | 680                    | 1024               |
-| YOLO26X_SEG | 1410                   | 1450               |
+| Model            | Android (XNNPACK) [MB] | iOS (XNNPACK) [MB] |
+| ---------------- | ---------------------- | ------------------ |
+| YOLO26N_SEG      | 92                     | 668                |
+| YOLO26S_SEG      | 220                    | 712                |
+| YOLO26M_SEG      | 570                    | 815                |
+| YOLO26L_SEG      | 680                    | 1024               |
+| YOLO26X_SEG      | 1410                   | 1450               |
+| RF_DETR_NANO_SEG | 620                    | 603                |
 
 ## Text to image
 

docs/docs/02-benchmarks/model-size.md

@@ -14,6 +14,17 @@ title: Model Size
 | ------------------------------ | :---------------: | :---------------: | :---------------: |
 | SSDLITE_320_MOBILENET_V3_LARGE |       13.9        |       15.6        |       8.46        |
 
+## Instance Segmentation
+
+| Model            | XNNPACK [MB] |
+| ---------------- | :----------: |
+| YOLO26N_SEG      |     11.6     |
+| YOLO26S_SEG      |     42.3     |
+| YOLO26M_SEG      |     95.4     |
+| YOLO26L_SEG      |     113      |
+| YOLO26X_SEG      |     252      |
+| RF_DETR_NANO_SEG |     124      |
+
 ## Style Transfer
 
 | Model                        | XNNPACK FP32 [MB] | XNNPACK INT8 [MB] | Core ML FP32 [MB] | Core ML FP16 [MB] |

docs/docs/03-hooks/02-computer-vision/useInstanceSegmentation.md

@@ -118,10 +118,11 @@ function App() {
 YOLO models use the [`CocoLabelYolo`](../../06-api-reference/enumerations/CocoLabelYolo.md) enum (80 classes, 0-indexed), which differs from [`CocoLabel`](../../06-api-reference/enumerations/CocoLabel.md) used by RF-DETR and SSDLite object detection models (91 classes, 1-indexed). When filtering with `classesOfInterest`, use the label names from `CocoLabelYolo`.
 :::
 
-| Model       | Number of classes | Class list                                                          | Available input sizes |
-| ----------- | ----------------- | ------------------------------------------------------------------- | --------------------- |
-| yolo26n-seg | 80                | [COCO (YOLO)](../../06-api-reference/enumerations/CocoLabelYolo.md) | 384, 512, 640         |
-| yolo26s-seg | 80                | [COCO (YOLO)](../../06-api-reference/enumerations/CocoLabelYolo.md) | 384, 512, 640         |
-| yolo26m-seg | 80                | [COCO (YOLO)](../../06-api-reference/enumerations/CocoLabelYolo.md) | 384, 512, 640         |
-| yolo26l-seg | 80                | [COCO (YOLO)](../../06-api-reference/enumerations/CocoLabelYolo.md) | 384, 512, 640         |
-| yolo26x-seg | 80                | [COCO (YOLO)](../../06-api-reference/enumerations/CocoLabelYolo.md) | 384, 512, 640         |
+| Model           | Number of classes | Class list                                                          | Available input sizes |
+| --------------- | ----------------- | ------------------------------------------------------------------- | --------------------- |
+| yolo26n-seg     | 80                | [COCO (YOLO)](../../06-api-reference/enumerations/CocoLabelYolo.md) | 384, 512, 640         |
+| yolo26s-seg     | 80                | [COCO (YOLO)](../../06-api-reference/enumerations/CocoLabelYolo.md) | 384, 512, 640         |
+| yolo26m-seg     | 80                | [COCO (YOLO)](../../06-api-reference/enumerations/CocoLabelYolo.md) | 384, 512, 640         |
+| yolo26l-seg     | 80                | [COCO (YOLO)](../../06-api-reference/enumerations/CocoLabelYolo.md) | 384, 512, 640         |
+| yolo26x-seg     | 80                | [COCO (YOLO)](../../06-api-reference/enumerations/CocoLabelYolo.md) | 384, 512, 640         |
+| rfdetr-nano-seg | 91                | [COCO](../../06-api-reference/enumerations/CocoLabel.md)            | N/A                   |

packages/react-native-executorch/common/rnexecutorch/models/VisionModel.h

@@ -90,25 +90,14 @@ class VisionModel : public BaseModel {
    */
   virtual cv::Mat preprocess(const cv::Mat &image) const;
 
-  /// Cached input tensor shape (getAllInputShapes()[0]).
-  /// Set once by each subclass constructor to avoid per-frame metadata lookups.
-  std::vector<int32_t> modelInputShape_;
-
   /**
    * @brief Get the spatial dimensions of the model input.
    *
    * By default, returns the last two dimensions of modelInputShape_.
    * Subclasses may override this for models with dynamic or multiple input
    * sizes.
    */
-  virtual cv::Size modelInputSize() const {
-    if (modelInputShape_.size() < 2) {
-      return {0, 0};
-    }
-    return cv::Size(modelInputShape_[modelInputShape_.size() - 1],
-                    modelInputShape_[modelInputShape_.size() - 2]);
-  }
-
+  virtual cv::Size modelInputSize() const;
   /**
    * @brief Extract an RGB cv::Mat from a VisionCamera frame
    *