Address tfjs-models typos in documentation strings

body-pix/README_Archive.md

@@ -141,7 +141,7 @@ const net = await bodyPix.load({
    - `2`. 2 bytes per float. Leads to slightly lower accuracy and 2x model size reduction.
    - `1`. 1 byte per float. Leads to lower accuracy and 4x model size reduction.
 
-   The following table contains the corresponding BodyPix 2.0 model checkpoint sizes (widthout gzip) when using different quantization bytes:
+   The following table contains the corresponding BodyPix 2.0 model checkpoint sizes (without gzip) when using different quantization bytes:
 
      | Architecture       | quantBytes=4 | quantBytes=2 | quantBytes=1 |
      | ------------------ |:------------:|:------------:|:------------:|

body-pix/src/setup_test.ts

@@ -27,7 +27,7 @@ import {setTestEnvs} from '@tensorflow/tfjs-core/dist/jasmine_util';
 // Increase test timeout since we are fetching the model files from GCS.
 jasmine.DEFAULT_TIMEOUT_INTERVAL = 20000;
 
-// Run browser tests againts both the cpu and webgl backends.
+// Run browser tests against both the cpu and webgl backends.
 setTestEnvs([
   // WebGL.
   {

body-segmentation/demos/shared/params.js

@@ -57,7 +57,7 @@ export const BLAZE_POSE_CONFIG = {
   visualization: 'binaryMask'
 };
 /**
- * This map descripes tunable flags and theior corresponding types.
+ * This map describes tunable flags and theior corresponding types.
  *
  * The flags (keys) in the map satisfy the following two conditions:
  * - Is tunable. For example, `IS_BROWSER` and `IS_CHROME` is not tunable,

body-segmentation/demos/shared/util.js

@@ -37,7 +37,7 @@ export function isMobile() {
 async function resetBackend(backendName) {
   const ENGINE = tf.engine();
   if (!(backendName in ENGINE.registryFactory)) {
-    throw new Error(`${backendName} backend is not registed.`);
+    throw new Error(`${backendName} backend is not registered.`);
   }
 
   if (backendName in ENGINE.registry) {

body-segmentation/src/body_pix/README.md

@@ -74,7 +74,7 @@ Pass in `bodySegmentation.SupportedModels.BodyPix` from the
    - `2`. 2 bytes per float. Leads to slightly lower accuracy and 2x model size reduction.
    - `1`. 1 byte per float. Leads to lower accuracy and 4x model size reduction.
 
-   The following table contains the corresponding BodyPix 2.0 model checkpoint sizes (widthout gzip) when using different quantization bytes:
+   The following table contains the corresponding BodyPix 2.0 model checkpoint sizes (without gzip) when using different quantization bytes:
 
      | Architecture       | quantBytes=4 | quantBytes=2 | quantBytes=1 |
      | ------------------ |:------------:|:------------:|:------------:|

body-segmentation/src/body_pix/impl/setup_test.ts

@@ -27,7 +27,7 @@ import { setTestEnvs } from '@tensorflow/tfjs-core/dist/jasmine_util';
 // Increase test timeout since we are fetching the model files from GCS.
 jasmine.DEFAULT_TIMEOUT_INTERVAL = 20000;
 
-// Run browser tests againts both the cpu and webgl backends.
+// Run browser tests against both the cpu and webgl backends.
 setTestEnvs([
   // WebGL.
   {

coco-ssd/src/index.ts

@@ -39,7 +39,7 @@ export interface DetectedObject {
  */
 export interface ModelConfig {
   /**
-   * It determines wich object detection architecture to load. The supported
+   * It determines which object detection architecture to load. The supported
    * architectures are: 'mobilenet_v1', 'mobilenet_v2' and 'lite_mobilenet_v2'.
    * It is default to 'lite_mobilenet_v2'.
    */
@@ -212,7 +212,7 @@ export class ObjectDetection {
 
   /**
    * Detect objects for an image returning a list of bounding boxes with
-   * assocated class and score.
+   * associated class and score.
    *
    * @param img The image to detect objects from. Can be a tensor or a DOM
    *     element image, video, or canvas.

deeplab/README.md

@@ -29,7 +29,7 @@ loadModel()
             console.log(`The predicted classes are ${JSON.stringify(legend)}`));
 ```
 
-By default, calling `load` initalizes the PASCAL variant of the model quantized to 2 bytes.
+By default, calling `load` initializes the PASCAL variant of the model quantized to 2 bytes.
 
 If you would rather load custom weights, you can pass the URL in the config instead:
 
@@ -136,7 +136,7 @@ const classify = async (image) => {
 
 ### Producing a Semantic Segmentation Map
 
-To segment an arbitrary image and generate a two-dimensional tensor with class labels assigned to each cell of the grid overlayed on the image (with the maximum number of cells on the side fixed to 513), use the `predict` method of the `SemanticSegmentation` object.
+To segment an arbitrary image and generate a two-dimensional tensor with class labels assigned to each cell of the grid overlaid on the image (with the maximum number of cells on the side fixed to 513), use the `predict` method of the `SemanticSegmentation` object.
 
 #### `model.predict(image)` input
 

deeplab/src/index.ts

@@ -112,7 +112,7 @@ export class SemanticSegmentation {
 
   /**
    * Segments an arbitrary image and generates a two-dimensional tensor with
-   * class labels assigned to each cell of the grid overlayed on the image ( the
+   * class labels assigned to each cell of the grid overlaid on the image ( the
    * maximum number of cells on the side is fixed to 513).
    *
    * @param input ::
@@ -133,7 +133,7 @@ export class SemanticSegmentation {
 
   /**
    * Segments an arbitrary image and generates a two-dimensional tensor with
-   * class labels assigned to each cell of the grid overlayed on the image ( the
+   * class labels assigned to each cell of the grid overlaid on the image ( the
    * maximum number of cells on the side is fixed to 513).
    *
    * @param image :: `ImageData | HTMLImageElement | HTMLCanvasElement |

depth-estimation/demos/depth_map/js/gl-class.js

@@ -29,7 +29,7 @@ class GlTextureImpl {
   }
 }
 
-// A wrapper class for WebGL texture and its associted framebuffer and utility
+// A wrapper class for WebGL texture and its associated framebuffer and utility
 // functions.
 class GlTextureFramebuffer extends GlTextureImpl {
   constructor(gl, framebuffer, texture, width, height) {

depth-estimation/demos/relighting/js/gl-class.js

@@ -29,7 +29,7 @@ class GlTextureImpl {
   }
 }
 
-// A wrapper class for WebGL texture and its associted framebuffer and utility
+// A wrapper class for WebGL texture and its associated framebuffer and utility
 // functions.
 class GlTextureFramebuffer extends GlTextureImpl {
   constructor(gl, framebuffer, texture, width, height) {

depth-estimation/demos/relighting/js/gl-shaders.js

@@ -47,7 +47,7 @@ out vec4 out_color;
 #define GetDepth(uv) (texture(uDepth, uv).r)
 #define GetColor(uv) (texture(uColor, uv).rgb)
 
-// Computes the aspect ratio for portait and landscape modes.
+// Computes the aspect ratio for portrait and landscape modes.
 vec2 CalculateAspectRatio(in vec2 size) {
   return pow(size.yy / size, vec2(step(size.x, size.y) * 2.0 - 1.0));
 }
@@ -177,7 +177,7 @@ vec3 RenderMotionLights(in vec2 uv) {
   col = smoothstep(0.0, 0.7, col + 0.05);
   col = pow(col, vec3(1.0 / 1.8));
 
-  // Perceptual light radius propotional to percentage in the screen space.
+  // Perceptual light radius proportional to percentage in the screen space.
   float light_radius = 2.0 * atan(kLightRadius, 2.0 * (1.0 - center.z));
 
   float l = distance(center.xy, normalized_uv);

depth-estimation/src/ar_portrait_depth/estimator.ts

@@ -128,7 +128,7 @@ class ARPortraitDepthEstimator implements DepthEstimator {
       // Shape after expansion is [1, height, width, 3].
       const batchInput = tf.expandDims(imageResized);
 
-      // Depth prediction (ouput shape is [1, height, width, 1]).
+      // Depth prediction (output shape is [1, height, width, 1]).
       const depth4D = this.estimatorModel.predict(batchInput) as tf.Tensor4D;
 
       // Normalize to user requirements.