Seldon QA Baseline — MarquiseRosier/Emotti

[seldonqa]

Seldon QA Baseline — MarquiseRosier/Emotti

Seldon analysed the default branch @ 9db78fb and generated 132 test cases.

Review & edit this plan in Seldon →

Set your target URL, include/exclude cases, edit steps, and run the plan from the Seldon QA editor. Once you approve it there, Seldon reviews every pull request automatically.

---

132 test cases | P0: 30 P1: 64 P2: 31 P3: 7

ID	Priority	Area	Test
TC-001	P0	Application Startup	Application Launches and Camera Feed Renders Without Errors (Smoke Test)
TC-002	P0	Data Augmentation – Batch Flow	Batch Generation (flow()): Train Mode Augmentation and Correct Batch Output Format
TC-003	P0	Data Augmentation – Random Crop	Random Crop Transform: Output Shape Validity and Bounding Box Handling Limitation
TC-004	P0	Data Augmentation – Random Rotation	Random Rotation Transform: Verify Actual Rotation vs. Copy-Paste Bug
TC-005	P0	Emotion Classification - End-to-End	Emotion Classification Pipeline Produces Valid Emotion Labels for All Supported States
TC-006	P0	Emotion Classification Model	Emotion classification model loads and produces valid 7-class probability output
TC-007	P0	Emotion Detection	Emotion Classification Pipeline: Webcam initializes and runs for 10-second capture window
TC-008	P0	Emotion Detection - CNN Inference	Emotion CNN Inference Returns Correct 7-Class Label Mapping
TC-009	P0	Emotion Detection - End-to-End	Emotion Detection End-to-End: Upload Image and Receive Emotion Label
TC-010	P0	Emotion Detection - Vote Accumulation &…	Dominant Emotion Selection via Voting Across 10-Second Window
TC-011	P0	Emotion Handler - State Routing	run_func() Correctly Routes Happy, Sad, and Neutral Emotions to Their Handlers
TC-012	P0	Emotion Inference	Emotion Model UI: Canonical Emotion Checkpoints Load and Produce 7-Class Output
TC-013	P0	Emotion Inference	End-to-End Emotion Detection Pipeline: Face Detected → Emotion Labeled in UI
TC-014	P0	Emotion Label Mapping	All 7 emotion labels are correctly identified and displayed by the application
TC-015	P0	Emotion Response - Angry	Angry Emotion Handler: Calming TTS message played and Google Images cute puppies page opens in new window
TC-016	P0	Emotion Response - Happy	Happy Emotion Handler: Compliment selected, TTS audio generated and played
TC-017	P0	Emotion Routing	Emotion Routing Logic: All 7 emotions route to correct handlers via run_func()
TC-018	P0	Emotion-Triggered Browser Action	Angry Emotion Trigger Opens Google Images Cute Puppies Page
TC-019	P0	End-to-End - Full Pipeline	Full End-to-End Emotion Detection and Response Flow (Happy Path)
TC-020	P0	End-to-End Flow	End-to-end emotion detection flow produces correct output for a detected face
TC-021	P0	Face Detection	Face Detection: Bounding Box Rendered Correctly on Detected Faces
TC-022	P0	Face Detection - Inference Pipeline	Face Detection Pipeline Detects Faces and Displays Bounding Boxes on Live Camera Feed
TC-023	P0	Grad-CAM Visualization	Grad-CAM Heatmap Visualization Renders for a Predicted Emotion Class
TC-024	P0	Model Inference / Emotion Classification	Emotion Model Inference — mini_XCEPTION Produces Valid Softmax Output (7 Classes)
TC-025	P0	Model Inference / Gender Classification	Gender Model Inference — mini_XCEPTION Produces Valid Softmax Output (2 Classes)
TC-026	P0	Per-Frame Face Detection	Face detection identifies faces and returns valid bounding box coordinates per frame
TC-027	P0	Sentiment Analysis (Azure Text Analytic…	Sentiment Analysis – Hardcoded Azure Subscription Key Exposure Check
TC-028	P0	UI / Emotion Detection	Emotion Detection UI — Displays Predicted Emotion Label on Face Input
TC-029	P0	UI / Gender Detection	Gender Detection UI — Displays Predicted Gender Label on Face Input
TC-030	P0	Webcam Capture Loop	Webcam capture loop runs for approximately 10 seconds and terminates cleanly
TC-031	P1	Browser Integration	Browser opens Google Images cute puppies search in new window on angry emotion
TC-032	P1	Browser Integration	Google Images cute puppies URL is accessible and returns valid image results
TC-033	P1	Compliment Resource Loading	Compliment JSON resource files exist, are valid, and contain non-empty string arrays
TC-034	P1	Data Augmentation – Batch Flow	Batch Generation (flow()): Validation Mode — No Augmentation Applied
TC-035	P1	Data Augmentation – Brightness	Brightness Transform: Pixel Values Stay Within [0, 255]
TC-036	P1	Data Augmentation – Contrast	Contrast Transform: Adjustment Around Mean Grayscale and Valid Output Range
TC-037	P1	Data Augmentation – Horizontal Flip	Horizontal Flip Transform: Image Reversal and Bounding Box Coordinate Update
TC-038	P1	Data Augmentation – Lighting	Lighting (PCA Noise) Transform: Numerical Stability and Output Clipping
TC-039	P1	Data Augmentation – Train/Val Split	Train/Validation Split: Mutual Exclusivity and No Data Leakage
TC-040	P1	Data Augmentation – Transform Pipeline	Color Jitter Pipeline (transform()): Random Order and Selective Application
TC-041	P1	Data Augmentation – Vertical Flip	Vertical Flip Transform: Image Reversal and Bounding Box Y-Coordinate Update
TC-042	P1	Dataset Utilities / Error Handling	Invalid dataset name raises a visible error in the application
TC-043	P1	Dataset Utilities / get_labels	fer2013 dataset label mappings are correct and complete
TC-044	P1	Emotion Detection	Emotion Classification Pipeline: Face detection handles multiple faces, no faces, and partial faces
TC-045	P1	Emotion Detection	Emotion Classification Pipeline: Graceful handling of missing or corrupted model file
TC-046	P1	Emotion Detection - Angry Handler	Angry Emotion Trigger Opens Google Images Cute Puppies Page
TC-047	P1	Emotion Detection - Camera Error Handli…	Camera Unavailability Causes Graceful Failure Without Application Crash
TC-048	P1	Emotion Detection - Edge Case / No Face	No Face Detected Defaults to 'Angry' Emotion (All-Zero Vote Array Edge Case)
TC-049	P1	Emotion Handler - Angry Dispatch	Angry Emotion Handler Opens Google Images Cute Puppies in New Browser Window
TC-050	P1	Emotion Handler - Angry Dispatch / URL …	Google Images Cute Puppies URL Returns Valid Image Search Results Page
TC-051	P1	Emotion Handler - Resource Loading	JSON Resource Files Load Correctly and Provide Valid Compliment Strings
TC-052	P1	Emotion Handler - State Routing	run_func() Routes Unmapped Emotions (surprise, disgust, fear) to Angry Handler
TC-053	P1	Emotion Handler - TTS & Audio Playback	gTTS Generates MP3 and mpg123 Plays Audio for Each Emotion Handler
TC-054	P1	Emotion Handling - Angry State	Angry Emotion Handler Opens Google Images Cute Puppies Page
TC-055	P1	Emotion Handling - Happy State	Happy Emotion Handler Produces a Non-Empty Compliment (Random Selection from comps.json)
TC-056	P1	Emotion Handling - Neutral State	Neutral Emotion Handler Produces a Non-Empty Anytime Compliment (Random Selection from compliments.json)
TC-057	P1	Emotion Inference	Emotion Label Mapping: Predictions Display Correct Human-Readable Labels
TC-058	P1	Emotion Response - Neutral	Neutral Emotion Handler: Random anytime compliment selected from compliments.json, TTS generated and played
TC-059	P1	Emotion Response - Sad	Sad Emotion Handler: Supportive TTS message generated and played; browser action remains disabled
TC-060	P1	Emotion Response - Unimplemented Handle…	Unimplemented Emotion Handlers (Surprise, Disgust, Fear): No crash and no user feedback produced
TC-061	P1	Emotion Response / UI Integration	Angry emotion trigger opens Google Images cute puppies search
TC-062	P1	Emotion Response / UI Integration	Angry Emotion Trigger Opens Google Images Cute Puppies Page
TC-063	P1	Emotion Routing	Emotion Routing Defect: Surprise, disgust, and fear incorrectly trigger angry() handler instead of their own …
TC-064	P1	Emotion-Triggered Browser Action	Google Images Cute Puppies URL Remains Accessible and Returns Valid Results
TC-065	P1	Error Handling / Path Resolution	Missing Emotion Model File: Application Surfaces Informative Error
TC-066	P1	Error Handling / Path Resolution	Missing Face Detection XML: Application Surfaces Informative Error
TC-067	P1	External URL Reachability	Google Images Cute Puppies URL Loads Successfully
TC-068	P1	Face Detection - Edge Cases	Face Detection Handles No-Face and Multi-Face Scenarios Gracefully
TC-069	P1	Face Detection - Inference Pipeline / U…	Emotion Label Text Is Rendered on Video Feed Alongside Bounding Box
TC-070	P1	Gender Inference	Gender Model UI: Canonical Gender Checkpoints Load and Produce Binary Output
TC-071	P1	Grad-CAM Visualization - Guided Backprop	Guided Grad-CAM Visualization Renders for Emotion and Gender Tasks
TC-072	P1	Image Preprocessing	Image preprocessing normalizes pixel values to [-1, 1] range and produces correct batch shape
TC-073	P1	Image Preprocessing - Grayscale Convers…	Grayscale Conversion Produces Correct Output Shape and Values
TC-074	P1	Image Preprocessing - Load Image	Image Load and Batch Dimension Expansion for Model Input
TC-075	P1	Image Preprocessing - Normalization	Image Preprocessing Normalization Applied in Train/Val Modes but Not Demo Mode
TC-076	P1	Image Preprocessing - preprocess_input	preprocess_input Normalization: v2=True Produces [-1,1] Range, v2=False Produces [0,1] Range
TC-077	P1	Image Preprocessing / load_image	Image loading from valid file path produces correct output
TC-078	P1	Image Preprocessing / load_image	Image loading with invalid or missing file path shows error
TC-079	P1	Model File Utilities	Model File Utilities – Optimizer Weights Removal from HDF5 File
TC-080	P1	Model Inference / Emotion Classification	Emotion Model Inference — simple_CNN Produces Valid Softmax Output (7 Classes)
TC-081	P1	Model Inference / Emotion Classification	big_XCEPTION Emotion Model Checkpoint Loads and Produces Valid Inference
TC-082	P1	Model Inference / Emotion Classificatio…	KDEF Emotion Model Inference — tiny_XCEPTION Checkpoint Loads and Predicts
TC-083	P1	Model Inference / Gender Classification	Gender Model Inference — simple_CNN Produces Valid 2-Class Output
TC-084	P1	Model Loading	Application handles missing Haar Cascade or emotion model files gracefully
TC-085	P1	Negative / Error Handling	No Face Input — Application Handles Gracefully Without Crash
TC-086	P1	Path Resolution / Configuration	Relative Path Resolution: Application Works When Launched from Project Root
TC-087	P1	Preprocessing - Input Normalization	Pixel Normalization Produces Correct Output Range for Neural Network Input
TC-088	P1	Sentiment Analysis (Azure Text Analytic…	Sentiment Analysis GUI – Text Input Capture and Commit Button Visibility
TC-089	P1	Sentiment Analysis (Azure Text Analytic…	Sentiment Analysis – API Request JSON Format Matches Azure Schema
TC-090	P1	Sentiment Analysis (Azure Text Analytic…	Sentiment Analysis – Empty Input Handling
TC-091	P1	Test Infrastructure	Baseline: Confirm absence of any test infrastructure in the repository
TC-092	P1	Text-to-Speech Audio Generation & Playb…	Text-to-Speech audio is generated and played for detected emotions
TC-093	P1	Word-Embedding Categoriser (GloVe Cosin…	Word-Embedding Categoriser – Empty String Query Edge Case
TC-094	P1	Word-Embedding Categoriser (GloVe Cosin…	Word-Embedding Categoriser – Hardcoded Demographic Category Bias Audit

38 more (P2/P3)

ID	Priority	Area	Test
TC-095	P2	Audio Playback	Application handles missing audio file gracefully when mpg123 attempts playback
TC-096	P2	Browser Integration	Emotion response flow is not blocked after browser window is opened for angry emotion
TC-097	P2	Data Augmentation – Batch Flow	Batch Generation (flow()): Demo Mode Applies Augmentation Without Preprocessing
TC-098	P2	Dataset Utilities / DataManager	DataManager initializes with correct default dataset paths
TC-099	P2	Dataset Utilities / get_class_to_arg	get_class_to_arg provides consistent round-trip mapping with get_labels for fer2013
TC-100	P2	Dataset Utilities / get_labels	IMDB dataset label mappings correctly identify gender classes
TC-101	P2	Dataset Utilities / get_labels	KDEF dataset emotion labels display correctly in prediction output
TC-102	P2	Dataset Utilities / split_data	Data split produces correct 80/20 train-validation ratio
TC-103	P2	Emotion Detection - Error Handling	Emotion Detection Resize Failure is Caught and Handled Gracefully
TC-104	P2	Emotion Detection – Angry Handler	Angry Emotion Handler: Google Images Cute Puppies Page Opens in Browser
TC-105	P2	Emotion Handling - Stub States	Stub Emotion Handlers (Surprise, Disgust, Fear) Execute Without Error
TC-106	P2	Emotion Response - Angry	Angry Emotion Handler: Google Images cute puppies URL is reachable and displays image results
TC-107	P2	External URL / Smoke	Google Images cute puppies page loads and displays image results
TC-108	P2	External URL Reachability	Google Images Cute Puppies Page Loads Successfully
TC-109	P2	Grad-CAM Visualization - Error Handling	Grad-CAM Invalid Layer Name Handling
TC-110	P2	Grad-CAM Visualization - Normalization	Grad-CAM Heatmap Normalization Clamps Negative Values and Scales to [0,1]
TC-111	P2	Image Preprocessing - Grayscale Convers…	Grayscale Conversion Applied Only When grayscale=True
TC-112	P2	Image Preprocessing / load_image	Grayscale image conversion produces single-channel output
TC-113	P2	Model File Utilities	Model File Utilities – Error Handling for Missing optimizer_weights Key
TC-114	P2	Model File Utilities	Model File Utilities – Error Handling for Invalid File Path
TC-115	P2	NLP / Sentiment Analysis Integration	Azure Text Analytics sentiment endpoint is reachable from the application
TC-116	P2	NLP / Sentiment Analysis Integration	Azure Sentiment API Endpoint Reachability Check
TC-117	P2	NLP / Sentiment Analysis Integration	Azure Text Analytics Sentiment API Endpoint Reachability
TC-118	P2	NLP – Sentiment Analysis	Sentiment Analysis API Endpoint: Reachability and Response Format
TC-119	P2	Negative / Edge Case	Multiple Faces in Frame — Application Handles Multi-Face Input
TC-120	P2	Preprocessing / Deprecated API	Deprecated scipy imread/imresize wrappers do not surface errors in production UI
TC-121	P2	Visualization - Draw Mosaic Error Handl…	Draw Mosaic Raises Error When class_vectors Provided Without Decoder
TC-122	P2	Visualization - Image Display Helpers	Image Display with Class Labels and Colorbar Renders Correctly
TC-123	P2	Visualization - Mosaic Grid	Image Mosaic Grid Renders Correctly in Visualization Output
TC-124	P2	Word-Embedding Categoriser (GloVe Cosin…	Word-Embedding Categoriser – Out-of-Vocabulary Word Handling
TC-125	P2	Word-Embedding Categoriser (GloVe Cosin…	Word-Embedding Categoriser – Category Score Determinism and Reproducibility
TC-126	P3	Emotion Handler – UI Navigation	Angry Emotion Handler – Google Images Cute Puppies Page Loads Correctly
TC-127	P3	Model Selection / Training Artifacts	Checkpoint Accuracy Progression: UI or Demo Reflects Best Model Selection
TC-128	P3	NLP / Sentiment Analysis	Azure Sentiment API Endpoint Accessibility Check
TC-129	P3	NLP Sentiment Analysis - API Connectivi…	Sentiment Analysis API Endpoint Reachability Check
TC-130	P3	Training Logs / Model Management	FER2013 Training Log — Verify Log File Is Accessible and Parseable via UI
TC-131	P3	Training Logs / Model Management	KDEF Training Log — Verify Lower Accuracy vs FER2013 Is Reflected in UI
TC-132	P3	Visualization / get_colors	Visualization color palette covers full spectrum with distinct colors

---

_{Generated by Seldon from MarquiseRosier/Emotti @ 9db78fb · Manage in Seldon}