Official curriculum roadmap, subject/topic distribution, negative marking rules, pacing guidelines, and solved sample questions.
🎯 Mapped Subjects & Topic Question Distribution
Total Question Pool100%
82 MCQs
Combined Active Syllabus
Computer Vision & Geospatial Remote Sensing AI
82 MCQs
Topic Pool
📊 Question Pool Structure
82 MCQs across fundamental, intermediate, and advanced concept tiers.
⚡ Recommended Pacing
45 to 60 seconds per MCQ. Flag complex problems and preserve 10 minutes for final revision.
⚖️ Scoring & Negative Marking
+1 mark per correct answer. In competitive tests with negative marking, -0.25 applies for incorrect guesses.
💡 Strategic Preparation & Exam Hall Guidelines
To maximize your score on Computer Vision & Geospatial Remote Sensing AI, candidates are advised to follow a structured three-pass approach. In the First Pass, solve all direct recall and formula-based questions within 30 seconds each to secure foundational marks. In the Second Pass, tackle multi-step analytical and quantitative reasoning problems. In the Third Pass, review marked questions and verify calculations.
Practice with the interactive player below to evaluate your speed and accuracy under real exam pressure. Every question features full mathematical formulas, step-by-step worked solutions, and conceptual explanations vetted by Apex Rankers Academy subject matter specialists.
Ready to test your knowledge?Launch interactive 1-by-1 practice with instant feedback, bookmarking, and step-by-step rationales.
Review the solved problems below to understand question phrasing, answer choices, and step-by-step solution logic prior to starting the full interactive practice drill:
What is the primary difference between Two-Stage and One-Stage Object Detectors?
ATwo-stage detectors (Faster R-CNN) first generate Region Proposals (RPN) and then classify/refine them; One-stage detectors (YOLO, SSD) predict bounding boxes and class probabilities directly in a single dense grid pass
BTwo-stage detectors process images in black and white; One-stage detectors in color
CTwo-stage detectors run on CPU; One-stage detectors on GPU
DTwo-stage detectors only detect 2 objects per image
✓ Correct Answer:A - Two-stage detectors (Faster R-CNN) first generate Region Proposals (RPN) and then classify/refine them; One-stage detectors (YOLO, SSD) predict bounding boxes and class probabilities directly in a single dense grid pass
📖 Step-by-Step Solution & Conceptual Rationale:
Two-stage models prioritize detection accuracy/mAP; one-stage models achieve real-time inference speeds (>60 FPS) suitable for video streams and drone feeds.
In 'Faster R-CNN', what is the role of the 'Region Proposal Network' (RPN)?
AA lightweight convolutional network that slides over feature maps to generate candidate bounding box proposals (RoIs) with objectness scores and anchor box offsets
BA module that scales image brightness
CA module that converts bounding boxes to text captions
DA classifier that performs OCR on road signs
✓ Correct Answer:A - A lightweight convolutional network that slides over feature maps to generate candidate bounding box proposals (RoIs) with objectness scores and anchor box offsets
📖 Step-by-Step Solution & Conceptual Rationale:
RPN replaced slow CPU selective search with fast GPU-based anchor proposals, sharing convolutional features with the downstream detection network.
In YOLO (You Only Look Once), what is the 'Non-Maximum Suppression' (NMS) post-processing algorithm?
AAn iterative algorithm that filters out overlapping redundant bounding boxes, keeping only the highest confidence box and eliminating boxes with IoU (Intersection over Union) > threshold
BA method to suppress dark background pixels in images
CA neural network layer that speeds up backpropagation
DA technique for cropping images to square shapes
✓ Correct Answer:A - An iterative algorithm that filters out overlapping redundant bounding boxes, keeping only the highest confidence box and eliminating boxes with IoU (Intersection over Union) > threshold
📖 Step-by-Step Solution & Conceptual Rationale:
NMS cleans up dense multi-box detections, ensuring each unique physical object is localized with exactly one bounding box.
What is 'Mean Average Precision' (mAP, e.g. mAP@0.5, mAP@0.5:0.95) in object detection?
AThe mean of the Average Precision (area under the Precision-Recall curve) calculated across all object categories and IoU threshold intervals
BThe average classification accuracy across all pixels
CThe speed of object detection in frames per second
DThe mean pixel intensity across RGB channels
✓ Correct Answer:A - The mean of the Average Precision (area under the Precision-Recall curve) calculated across all object categories and IoU threshold intervals
📖 Step-by-Step Solution & Conceptual Rationale:
COCO mAP@0.5:0.95 averages AP across 10 IoU thresholds (from 0.50 to 0.95 with step 0.05) to rigorously penalize localization inaccuracies.
What is 'U-Net' architecture (Ronneberger et al. 2015) and why is it the gold standard for biomedical and satellite image segmentation?
AAn encoder-decoder network featuring skip connections that concatenate high-resolution spatial feature maps from contracting paths directly to expanding upsampling paths
BA network shaped like a circle that loops infinitely
CA generative network that produces synthetic images from text
DA network that processes 1D audio waves
✓ Correct Answer:A - An encoder-decoder network featuring skip connections that concatenate high-resolution spatial feature maps from contracting paths directly to expanding upsampling paths
📖 Step-by-Step Solution & Conceptual Rationale:
U-Net skip connections preserve precise pixel-level spatial boundaries lost during downsampling, essential for flood inundation and tumor boundary segmentation.
What is the primary difference between 'Semantic Segmentation' and 'Instance Segmentation'?
ASemantic segmentation labels every pixel with a class (e.g. all people are red); Instance segmentation labels every pixel AND differentiates distinct object instances (e.g. Person 1 is red, Person 2 is blue)
BSemantic is for text; Instance is for images
CSemantic is 2D; Instance is 3D
DThere is no difference between them
✓ Correct Answer:A - Semantic segmentation labels every pixel with a class (e.g. all people are red); Instance segmentation labels every pixel AND differentiates distinct object instances (e.g. Person 1 is red, Person 2 is blue)
📖 Step-by-Step Solution & Conceptual Rationale:
Instance segmentation (e.g. Mask R-CNN, SAM) detects individual objects with distinct mask boundaries, whereas semantic segmentation (e.g. DeepLab) groups all instances into one class mask.
What is 'Vision Transformer' (ViT, Dosovitskiy et al. 2020)?
AAn architecture that splits an image into a sequence of non-overlapping patches (e.g. 16x16 pixels), projects them linearly into token embeddings, and processes them with standard Transformer Encoders
BA transformer used in optical fiber internet cables
CA camera lens that zooms automatically using AI
DA convolutional network that processes text prompts
✓ Correct Answer:A - An architecture that splits an image into a sequence of non-overlapping patches (e.g. 16x16 pixels), projects them linearly into token embeddings, and processes them with standard Transformer Encoders
📖 Step-by-Step Solution & Conceptual Rationale:
ViT showed that standard Transformer architectures with minimal inductive bias surpass CNNs on large pre-training scales (JFT-300M, ImageNet-21k).
What is 'Swin Transformer' (Hierarchical Vision Transformer using Shifted Windows)?
AA vision transformer that computes self-attention within local non-overlapping windows and shifts window partitioning across layers to achieve linear O(N) computational complexity
BA transformer that swings back and forth during training
CA vision model trained on swimming pool images
DA model that processes images in reverse order
✓ Correct Answer:A - A vision transformer that computes self-attention within local non-overlapping windows and shifts window partitioning across layers to achieve linear O(N) computational complexity
📖 Step-by-Step Solution & Conceptual Rationale:
Swin Transformer provides multi-scale hierarchical feature maps and linear computational complexity relative to image size, making it a powerful general backbone for dense prediction.
Practice All 82 Questions InteractivelyTest your knowledge in real-time with continuous progress saving, instant scoring, and performance analytics.
Loading questions from the question bank...
No questions found matching your selected filters.